Myocontrolled robotic hands require accurate and responsive control to regulate grasp strength effectively. However, many human-in-the-loop (HITL) control systems still lack robust closed-loop solutions for fine grip force regulation, limiting their performance. This paper presents a novel control system for myocontrolled hands that combines contact force sensing and vibrotactile feedback to enable more natural and precise grasp interaction. The system features an advanced force controller based on fuzzy logic, with parameter optimization guided by user preferences collected through a graphical user interface (GUI) using Global Learning of Input–Output Strategies from Pairwise Preferences (GLISp). It is compared against heuristic model and neural network based controllers. The system was validated through real-world experiments using the AR10 robotic hand with OptoForce fingertip sensors, demonstrating improved adaptability and fine force regulation capabilities for the user.

Dual-arm manipulation poses many challenges in order to actually perform manipulation tasks with enough efficiency. To cope with this challenges, this work-in-progress describes a system for motion coordination of a mobile anthropomorphic dual-arm robot that focuses on the capability for performing manipulations actions with precision by means of a visual guidance provided by a camera mounted on an articulated head. The proposal presents different control modes, evaluated with real experiments, and discusses the main future research lines.

We introduce a simple yet effective approach for dense depth reconstruction that operates directly on raw disparity data, eliminating the need for additional disparity refinement stages. By leveraging disparity maps generated from conventional stereo methods, we train a U-Net-based model to directly map disparity to depth, bypassing complex feature engineering. Our method capitalizes on the robustness of traditional stereo matching techniques to varying scenes, focusing exclusively on dense depth reconstruction. This approach not only simplifies the training process but also significantly reduces the requirement for large-scale training datasets. Extensive evaluations demonstrate that our method surpasses classical stereo matching frameworks and state-of-the-art classical post-refinement techniques, achieving superior accuracy. Additionally, our approach offers competitive inference times, comparable to classical as well as end-to-end deep learning methods, making it highly suitable for real-time robotic applications.

Mobile manipulators working in semi-structured human environments need smart vision capabilities to perceive the world and interact with human operators. With this purpose in mind, this paper presents the development of a robotic head module composed of an OAK-D camera mounted on a pan-and-tilt WidowX XM430 structure. The OAK-D camera provides high-resolution images, including stereo vision and depth sensing, and advanced capabilities based on embedded artificial intelligence functions. These features combined with the high-range motion of the WidowX XM430, allow the head module to have advanced visual tracking capabilities. The implementation has been done using ROS (Robot Operating System), which allows the head module to be easily integrated to any mobile manipulator.

This work introduces a fully automatic and adaptable pipeline for synthetic depth dataset generation that later on can be used for the training of deep learning algorithms for robotic manipulation. From any available set of 3D object model meshes, the pipeline outputs rendered depth image data with labeled grasp candidates represented as a set of grasping points relative to the camera with associated metrics. The proposed pipeline allows adaptability in various characteristics such as the input dataset of objects, the sampling method, the gripper type, or the grasp evaluation metrics to allow the generation of a more customized, task-oriented collection of labeled grasps relevant for different robotic applications. The implementation is done using Blender’s Python API workspace, avoiding the use of multiple software tools or libraries, reducing the pipeline complexity, facilitating the extensibility, and providing benefits in terms of visualization and debugging.

Autonomous service robots are conceived to work in semi-structured and complex human environments performing a wide range of tasks and, hence, one of their main challenges is to be able to adapt the stages of the perceive-plan-execute cycle to perturbations ranging from small deviations on the poses of objects to large unexpected changes in the environment, as well as to recover from potential failures. To advance in this direction, this paper proposes an ontology-based manipulation framework where reasoning is used to enhance perception with situation awareness, planning with domain awareness and execution with the awareness of the execution structures. The combination of these different types of awareness allows the robot to have different adaptation capabilities. The conceptual schema of the framework is presented and discussed and the main future implementation challenges are pointed out.

The article deals with the problem of planning in the task space in the presence of vector fields, while verifying and validating the constraints in the configuration space. The proposed approach, called the Task Space Vector Field Rapidly-exploring Random Tree (TSVF-RRT) algorithm, extends the Task-Space Rapidly-exploring Random Trees (TS-RRT) algorithm by incorporating vector fields into the task space, while avoiding non-trivial constraints on the configuration space. The planner restricts the search to a lower dimensional space, minimizing the upstream functional. To evaluate the proposed approach, graphical simulations are presented, carried out using a planar manipulator of 10 DOF. Possible advantages that encourage further research in this line are discussed.

This paper describes a procedure to generate valid robot trajectories for a given sequence of points defining a geometric path, which is a quite frequent output of many robot motion planners. The proposed approach takes into account physical constraints of the robot, as the maximum velocity and the maximum acceleration that each joint can reach, and has been implemented in C++ as a general tool that can be used with different robots.

The execution of robotic manipulation tasks needs to be robust in front of failures or changes in the environment, and for this purpose, Behavior Trees (BT) are a good alternative to Finite State Machines, because the ability of BTs to be edited during run time and the fact that one can design reactive systems with BTs, makes the BT executor a robust execution manager. However, the good monitoring of the system state is required in order to react to errors at either geometric or symbolic level requiring, respectively, replanning at motion or at task level. This paper make a proposal in this line and, moreover, makes task planning adaptive to the actual situations encountered by knowledge-based reasoning procedures to automatically generate the Planning Domain Definition Language (PDDL) files that define the task.

Industry 4.0 demands interconnected production lines that consist of modular assets. Recent advances of wireless communication technologies allow a large connectivity of devices and approach the performance of wireline communication, specifically regarding throughput, latency and reliability. As a result, more and more time critical connections can be performed wirelessly. Both attributes foster the emergence of edge computing, a concept that can efficiently utilize distributed computational resources. This is particularly beneficial for modular assets that have limited energy supply and capacity of computation hardware. Autonomous mobile robots offer high potential for object transportation, inspection and manipulation in shared workspaces with human operators. With edge computing, heavy computations can then be offloaded to more powerful computers or edge data centers to speed up the decision-making process and increase the productivity. For an efficient orchestration strategy of computation and communication resources, various task requirements in terms of latency, bandwidth, cost and energy must be considered. To this end, we aim at evaluating the requirements in autonomous and collaborative assembly lines, a use case that comprises diverse tasks including latency-sensitive ones in dynamic, uncertain, multi-agent environments. This work focuses on discussing latency requirements on the basis of a collaborative safety mode and autonomous robotic insertion.

Learning-based approaches for motion planning have gained much attention. These approaches show better performance in terms of computational efficiency as compared to the tradition approaches, such as sampling-based motion planning. One of the key challenges for learning-based approaches is to generate training datasets. This work-in-progress proposes a generalized dataset generation approach for motion planning that allows to generate simulation data for motion planning under geometric, kinodynamic and physics-based constraints. The generated dataset can easily be encoded in any required format for training. This tool will be used at the Institute of Industrial and Control Engineering (IOC-UPC) to test learning-based motion planning strategies for dual arm mobile manipulators under physics-based constraints. As a first step, in this study the developed approach is tested by generating datasets for mobile robots. The generated datasets are used to train motion planning networks, an approach for deep learning-based motion planning.

Robots should be able to exercise reasoning in both symbolic and geometric levels in order to plan a manipulation task. The execution of such tasks needs to be robust enough to cope with real environments. In an attempt to address this pertinent industry need, the paper proposes the use of behavior trees for effective robotic manipulation in dynamic environments. This paper presents a method to automatically generate a behavior tree and showcases its ability to enable the robot to reason at different levels and adapt to an uncertain and changing environment. This allows for a complex task to be robustly executed, pioneering the advancement towards fully functional service robots.

Sensing is an essential component for robots to perform the manipulation tasks in real environments. This study proposes a lightweight deep-learning-based sensing modules which allows the robots to automatically model the workspace for manipulation planning. This sensing module is developed as a part of our ongoing manipulation planning framework. It will be used to enhance the sensing accuracy and make it capable of planning the manipulation tasks in real environments. The retrained model is further trained over commonly used objects to enhance the prediction accuracy.

Robotic manipulation requires the planning at a symbolic level (task planning) and at a geometric level (motion planning). This paper presents a planning framework for both levels of planning that includes an easy way to configure their interconnection. Motion planning is done using The Kautham Project, which is equipped with the Open Motion Planning Library suite of sampling-based motion planners, and task planning is done using the Fast Forward task planner. Both planning levels can be accessed through Robotic Operating System interfaces using services. A client program then uses these task and motion planning services and an XML configuration file that defines the linkage between symbolic actions and geometric values, to compute the sequence of feasible robot motions that allow to successfully execute a manipulation task. An illustrative example using the TIAGo mobile manipulator in a kitchen environment is presented where the flexibility in configuring different instances of manipulation tasks is shown.

Complex manipulation tasks require grasping strategies that simultaneously satisfy the stability and the semantic constraints that have to be satisfied for an action to be feasible, referred as action-oriented semantic grasp strategies. This study develops a framework using machine learning techniques to compute action-oriented semantic grasps. It takes a 3D model of the object and the action to be performed as input and provides a vector of action-oriented semantic grasps. We evaluate the performance of machine learning (particularly classification techniques) to determine which approaches perform better for this problem. Using the best approaches, a multi-model classification technique is developed. The proposed approach is evaluated in simulation to grasp different kitchen objects using a parallel gripper. The results show that multi-model classification approach enhances the prediction accuracy. The implemented system can be used as to automate the data labeling process required for deep learning approaches.

Robotic systems composed of several mobile robots moving in human environments pose several problems at perception, planning and control levels. In these environments, there may be obstacles obstructing the paths, which robots can remove by pushing or pulling them. At planning level, therefore, an efficient combination of task and motion planning is required. Even more if we assume a cooperative system in which robots can collaborate with each other by e.g. pushing together a heavy obstacle or by one robot clearing the way to another one. In this paper, we cope with this problem by proposing κ-TMP, a smart combination of an heuristic task planner based on the Fast Forward method, a physics-based motion planner, and reasoning processes over the ontologies that code the knowledge on the problem. The significance of the proposal relies on how geometric and physics information is used within the computation of the heuristics in order to guide the symbolic search, i.e. how an artificial intelligence planning method is combined with low-level motion planning to achieve a feasible sequence of actions (composed of collision-free motions plus physically-feasible push/pull actions). The proposal has been validated with several simulated scenarios (using up to five robots that need to collaborate with each other to reach the goal state), showing how the method is able to solve challenging situations and also find an efficient solution in terms of power.

This paper presents a planning procedure that allows an anthropomorphic dual-arm robotic system to perform a manipulation task in a natural human-like way by using demonstrated human movements. The key idea of the proposal is to convert the demonstrated trajectories into attractive potential fields defined over the configuration space and then use an RRT*-based planning algorithm that minimizes a path-cost function designed to bias the tree growth toward the human-demonstrated configurations. This paper presents a description of the proposed approach as well as results from a conceptual and a real application example, the latter using a real anthropomorphic dual-arm robotic system. A path-quality measure, based on first-order synergies (correlations between joint velocities) obtained from real human movements, is also proposed and used for evaluation and comparison purposes. The obtained results show that the paths obtained with the proposed procedure are more human-like.

This paper addresses the problem of obtaining human-like motions on hand-arm robotic systems performing grasping actions. The focus is set on the coordinated movements of the robotic arm and the anthropomorphic mechanical hand, with which the arm is equipped. For this, human movements performing different grasps are captured and mapped to the robot in order to compute the human hand synergies. These synergies are used to both obtain human-like movements and to reduce the complexity of the planning phase by reducing the dimension of the search space. In addition, the paper proposes a sampling-based planner, which guides the motion planning following the synergies and considering different types of grasps. The introduced approach is tested in an application example and thoroughly compared with a state-of-the-art planning algorithm, obtaining better results.

This paper discusses motion planning for anthropomorphic dual-arm robots. It introduces a measure of the similarity of the movements needed to solve two given tasks. Planning using this measure to select proper arm synergies for a given task improves the planning performance and the resulting plan.

In order to solve mobile manipulation problems, the efficient combination of task and motion planning is usually required. Moreover, the incorporation of physics-based information has recently been taken into account in order to plan the tasks in a more realistic way. In the present paper, a task and motion planning framework is proposed based on a modified version of the Fast-Forward task planner that is guided by physics-based knowledge. The proposal uses manipulation knowledge for reasoning on symbolic literals (both in offline and online modes) taking into account geometric information in order to evaluate the applicability as well as feasibility of actions while evaluating the heuristic cost. It results in an efficient search of the state space and in the obtention of low-cost physically-feasible plans. The proposal has been implemented and is illustrated with a manipulation problem consisting of a mobile robot and some fixed and manipulatable objects.

For everyday manipulation tasks, the combination of task and motion planning is required regarding the need of providing the set of possible subtasks which have to be done and how to perform them. Since many alternative plans may exist, the determination of their feasibility and the identification of the best one is a great challenge in robotics. To address this, this paper proposes: a) a version of GraphPlan (one of the best current approaches to task planning) that has been modified to use ontological knowledge and to allow the retrieval of all possible plans; and b) a physics-based reasoning process that determines the feasibility of the resulting plans and an associated cost that allows to select the best one among them. The proposed framework has been implemented and is illustrated through an example.

The paper deals with the problem of designing an RRT*-based planning algorithm that allows the user to guide the tree growth in a simple and transparent way. The key idea of the proposal is to create a planning algorithm, called HG-RRT*, that minimizes an optimization function over the configuration space where a state cost function is established. This state cost is defined as the combination of several potential fields. Each of these potential fields will attract the solution path or move it away from certain areas. The planning algorithm will try to minimize the path length, the motion effort and the variations of the cost along the path. The paper presents a description of the proposed approach as well as simulation results from a conceptual and an application example, including a thorough comparison with the TRRT planning algorithm.

Robotic manipulation involves actions where contacts occur between the robot and the objects. In this scope, the availability of physics-based engines allows motion planners to comprise dynamics between rigid bodies, which is necessary for planning this type of actions. However, physics-based motion planning is computationally intensive due to the high dimensionality of the state space and the need to work with a low integration step to find accurate solutions. On the other hand, manipulation actions change the environment and conditions further actions and motions. To cope with this issue, the representation of manipulation actions using ontologies enables a semantic-based inference processe that alleviates the computational cost of motion planning. This paper proposes a manipulation planning framework where physics-based motion planning is enhanced with ontological knowledge representation and reasoning. The proposal has been implemented and is illustrated and validated with a simple example. Its use in grasping tasks in cluttered environments is currently under development.

The paper deals with the problem of planning movements of dual-arm anthropomorphic systems, with the aim of reducing the computational cost of the problem and making the movements look as human-like as possible. The key idea of the proposal is the search of synergies of the dual-arm anthropomorphic system in order to use them to reduce the dimension of the search space while preserving human-like appearance. This idea was already developed and successfully used to plan movements of robotic hands, thus the extension to a dual-arm system is attractive. The paper presents a description of the proposed approach as well as real experimental results that encourage doing further research in this line.

This paper proposes a novel motion planning approach that exploits the concept of synergies (correlations) between degrees of freedom, extending it to the velocity space and calling them first-order synergies. An automatic partition method is defined to optimally divide the configuration space into cells where first-order synergies are significantly different. Using this partition, an algorithm that tends to grow a tree by extending the branches in the directions determined by the the first-order synergies of the cell where the leaf to be grown lies is introduced and called FOS-RRT. This allows the natural expansion of the tree along the directions determined by the data used to define the synergies. 2D examples illustrate the performance of the proposed approach, which is particularly attractive for potential applications in human-like robots using human synergies.

This paper presents the software tool used at the Institute of Industrial and Control Engineering (IOC-UPC) for teaching and research in robot motion planning. The tool allows to cope with problems with one or more robots, being a generic robot defined as a kinematic tree with a mobile base, i.e. the tool can plan and simulate from simple two degrees of freedom free-flying robots to multi-robot scenarios with mobile manipulators equipped with anthropomorphic hands. The main core of planners is provided by the Open Motion Planning Library (OMPL). Different basic planners can be flexibly used and parameterized, allowing students to gain insight into the different planning algorithms. Among the advanced features the tool allows to easily define the coupling between degrees of freedom, the dynamic simulation and the integration with task planers. It is principally being used in the research of motion planning strategies for hand-arm robotic systems.

The planning of collision-free paths of a team of mobile robots involves many degrees of freedom and therefore the use of sampling-based methods is a good alternative. Among them, the RRT∗ planner has been proposed to cope with optimization problems, and has been proven to be asymptotically optimal. Any optimization function can be defined, although optimization has been usually focused on the traveled distance or on safety, i.e. to find paths of minimum length or maximum clearance. Other constraints to be considered are related to the coordinate movements of the robots, including aspects like keeping a desired formation o having some similar behavior. In this paper we propose the use of an RRT∗ to optimize the traveled distance but subject to a coupled behavior between robots, i.e. it is desired that the robots behave as a group with similar or coordinated movements. To achieve so, a cost function has been defined that evaluates the alignment of the edges of the RRT∗ with the vectors that define the coupling between the motion directions of the robots. The method establishes a compromise between the independence required to avoid obstacles in a flexible way and the desired coupling to behave as a team. The method is illustrated with several examples.

Teleoperating a robotic hand with the aid of a sensorized glove presents some particular problems. A certain problem is due to the kinematic differences between the human hand and the robotic hand, which do not allow a simple direct mapping of the sensor readings from the glove to the robotic hand. This problem is addressed with different types of mapping, but none of them is of general use. This paper proposes two new mappings within two existing mapping types, as well as a new hybrid mapping that combines the best features of these existing mapping types. This hybrid mapping allows intuitive free space movements (where the gesture is more important than the precise positions of the fingers) and grasp movements (where the precise positions of the fingers is more important than the gesture), despite kinematic differences between the human hand and the robotic hand. The approach has been implemented, and some illustrative examples are presented in this paper.

Bronchoscopy is an interventional medical procedure employed to analyze the interior side of the human airways, clear possible obstructions and biopsy. Using a 3D reconstruction of the tracheobronchial tree, Virtual Bronchoscopy (VB) may help physicians in the exploration of peripheral lung lesions. We are developing a haptic-based navigation system for the VB that allows the navigation within the airways using a haptic device whose permitted motions mimics those done with the real bronchoscope. This paper describes the motion planning module of the system devoted to plan a path from the trachea to small peripheral pulmonary lesions, that takes into account the geometry and the kinematic constraints of the bronchoscope. The motion planner output is used to visually and haptically guide the navigation during the virtual exploration using the haptic device. Moreover, physicians can get useful information of whether the peripheral lesions can effectively be reached with a given bronchoscope or of which is the nearest point to the lesion that can be reached.

Teleoperation tasks are performed at cartesian level when the robot and the haptic device have dissimilar kinematics. If the size of the workspaces is also dissimilar, as it is usually the case, the mapping between workspaces must be handled with care in order to let the user teleoperate the robot in a natural and precise way. This paper formulates the mapping of workspaces based on the re-synchronization method and proposes an assisted system that lightens the user from the tedious part of the method, by guiding him/her towards the best re-synchronization position, thus minimizing the number of jumps. The proposal is part of a teleoperated guiding system being developed by the authors.

Path planning methods are well suited to automatically perform robotic tasks; haptic guidance is a powerful tool for disabled people rehabilitation, sports training, handcraft skills acquiring and such kind of enactive tasks. This paper proposes a novel an efficient method to accomplish guided movements by efficiently combing path planning methods and haptic guidance. The main contribution of this paper is the development of a reliable method to haptically guide the user within a virtual robotic task, by means of: on-line, collision free, path planning generated trajectories. To accomplish haptic guidance three stages must be correctly done: the first one where the user selects the desired objects and obstacles within a virtual environment to set up the task, the second one, where a path planner looks for the task trajectory and, the third one, where the user is constrained to on-line generated local channels and solution paths that offer a reliable path guidance system to achieve the task movements.

One of the key factors that affect the success and efficiency of sampling-based path planners is the obtention of samples in the more relevant regions of the workspace. This is known as importance sampling, and different approaches have already been proposed in this direction. This paper proposes a novel method to bias sampling by means of geometric constraints that reduces the sampling space to sets of lower dimensional submanifolds. These constraints may be imposed by the kinematic structure of the actuation system, by the task specification, or provided by a human user as an intuitive way to include problem knowledge to the planner. The method has been implemented and tested on a probabilistic roadmap planner giving promising results. A variant using a deterministic sampling source is also reported.

The paper deals with the problem of motion planning of anthropomorphic mechanical hands avoiding collisions. The proposed approach tries to mimic the real human hand motions, but reducing the dimension of the search space in order to obtain results as a compromise between motion optimality and planning complexity (time) by means of the concept of principal motion directions. Basically, the work includes the following phases: capturing the human hand workspace using a sensorized glove and mapping it to the mechanical hand workspace, reducing the space dimension by looking for the most relevant principal motion directions, and planning the hand movements using a sampling-based roadmap planner. The approach has been implemented for a four finger anthropomorphic mechanical hand, and some examples are included to illustrate its validity.

The planning of collision-free motions of a hand-arm system to reach a grasp or preshape configuration is not a simple issue due to the high number of involved degrees of freedom. This paper presents an efficient sampling-based path planner that copes with this issue by considering a reduced search space. The dimension of this space is not fixed but it is iteratively increased according to the difficulty of the task at hand. Initially the search space is 1-dimensional along the line defined by the initial and goal hand configurations (by construction those configurations always belong to the search space), and then its dimension is increased by iteratively adding principal motion directions (that couple the finger motions), trying in this way to produce hand movements through anthropomorphic natural postures.

This paper presents the use of independent contact and non-graspable regions to generate the grasp space for 2D and 3D discrete objects. The grasp space is constructed via a sampling method, which provides samples of force-closure or non force-closure grasps, used to compute regions of the graspable or non-graspable space, respectively. The method provides a reliable procedure for an efficient generation of the whole grasp space for n-finger grasps on discrete objects; two examples on 2D objects are provided to illustrate its performance. The approach has several applications in manipulation and regrasping of objects, as it provides a large number of force-closure and non force-closure grasps in a short time.

Hierarchical cell decompositions of Configuration Space can be of great value for enhancing sampling-based path planners, as well as for other robotic tasks with requirements beyond the planning of free paths. This paper proposes an efficient method to obtain a hierarchical cell decomposition of C-space that is based on: a) the use of a deterministic sampling sequence that allows an uniform and incremental exploration of the space, and b) the use of distance measurements to handle as much information as possible from each sample in order to make the procedure more efficient. The proposed cell decomposition procedure is applied to different path planning methods.

Sampling-based path planners are giving very good results for problems with high degrees of freedom, being the obtention of samples a crucial factor in their performance. Sampling generation is a difficult issue that entails the challenge of obtaining an incremental and uniform coverage of the configuration space. This objective is even harder when the planning of motions of 3D rigid bodies is tackled, since the corresponding configuration space is SE(3), which requires the careful handling of rotations. This paper proposes a deterministic sampling sequence that generates configurations of SE(3) for the planning of 3D rigid-body motions using sampling-based methods. The proposed sequence is able to generate samples in an incremental low-dispersion manner, giving a good uniform coverage and a lattice structure.

Potential-field approaches based on harmonic functions have good path planning properties, although the explicit knowledge of the robot’s Configuration Space is required. To overcome this drawback, a combination with a random sampling scheme is proposed. Harmonic functions are computed over a 2 d-tree decomposition of a d-dimensional Configuration Space that is obtained with a probabilistic cell decomposition (sampling and classification). Cell sampling is biased towards the more promising regions by using the harmonic function values. Cell classification is performed by evaluating a set of configurations of the cell obtained with a deterministic sampling sequence that provides a good uniform and incremental coverage of the cell. The proposed planning framework opens the use of harmonic functions to higher dimensional C-spaces.

Given an industrial robot equipped with a dexterous hand and an object to be grasped with four grasping points determined on its faces, this paper deals with the problem of finding the joint configurations that allow to grasp that object. The proposed solution is based on an iterative optimization method that consecutively moves the joint that best contributes to reduce the distance of the fingertips to the desired locations. The method is particularized for a Stäubli RX90 robot and the dexterous hand MA-I with four fingers developed at the IOC’s Robotics Lab.

This paper presents a new approach for the haptic rendering of compliant motions during the execution of virtual assembly tasks between simple polyhedral objects. The method, based on the task configuration space, analyzes the type of contacts that take place between objects, since this knowledge allows a better haptic rendering when face-face or edge-face contacts occur. Making use of spacial and temporal coherence, the paper presents an efficient procedure to keep track of the current contacts. This allows to comply to the hard temporal constraints of the haptic servo loop. The presented procedures are focused on the haptic rendering during the interaction between convex polyhedra.

This paper presents a robot motion planning method, called PHM that uses a random sampling scheme together with a potential-field approach based on harmonic functions. The combination of both results in an efficient path planner that is both resolution and probabilistic complete. On one hand, random sampling allows the use of the harmonic functions approach without the explicit knowledge of the robot's configuration space. On the other hand, harmonic functions allow an intelligent sampling of configuration space by introducing a bias towards the more promising regions.

Robot path planning copes with the planning of a collision-free path for the robot from an initial configuration to a goal configuration. The potential field method is one of the main approaches to path planning, and the use of harmonic functions allows the definition of potential fields without local minima and with the ability to be incrementally updated. Constrained motion planning copes with the planning of motions in cluttered environments with low clearances and high uncertainties. The robust, adaptive and reactive behavior of the harmonic functions approach suits the requirements of constrained motion planning. To efficiently compute harmonic functions for this purpose, the paper presents a hierarchical and dynamic method that works on a non-regular grid decomposition of configuration space.

The performance of tasks with robots in environments with low clearances (e.g. robotized assembly) is usually difficult, due mainly to geometric uncertainty and tolerances. Following a pragmatical approach to planar constrained-motion planning, this paper proposes a method for checking the feasibility of paths generated by gross-motion planning algorithms, taking into account the uncertainties affecting the task and the use of a compliant control mode. The method enables the extension of gross-motion planning techniques to constrained-motion planning problems, ensuring the feasibility of the task despite the uncertainties.

This paper presents a method for the planning and execution of compliant motions within the scope of a two-phase fine-motion planner for the performance of planar assembly tasks with robots. Algorithms are provided to find a nominal solution path in both free and contact configuration space which is feasible in spite of the uncertainties affecting the task. Compliant-motion commands based on the generalized damping control mode are synthesized to follow this path, allowing to maintain a constant bounded force.

This paper presents the graphical embedding of the rotational degree of freedom into the translational configuration space for planar assembly tasks. The resulting representation helps the visualization and understanding of motions in a simple and easy graphical way. The aim is the developing of motion planning algorithms in a two-dimensional parametrized space for planar assembly tasks (3 DOF) and in a three-dimensional parametrized space for assembly tasks in space (6 DOF). The method has been successfully applied to the problem of contact identification in the presence of uncertainty for planar assembly tasks. The extension to six degrees of freedom is under development.

Force-compliant control is an important aspect for performing assembly tasks with robots, since the geometric constraints of the task are usually used as guides. The paper presents the specification of compliant motions for planar assembly tasks (two degrees of freedom of translation and one of rotation), taking into account the effect of friction.