The papers included in this special section were presented at the Sixth Annual ACM/IEEE International Conference on Cyber-Physical Systems (ICCPS 2015) that was held on April 14-16, 2015 in Seattle, WA, USA, as part of the Eighth Annual Cyber-Physical Systems Week. ICCPS is the premier single-track conference for reporting advances in all aspects of cyber-physical systems, including theory, tools, applications, systems, testbeds, and field deployments. Its focus includes the core science and technology for developing fundamental principles that underpin the integration of cyber and physical elements, with application domains that include transportation, energy, water, agriculture, ecology, supply-chains, medical and assistive technology, sensor and social networks, and robotics.

High-level spinal cord injury (SCI) in humans causes paralysis below the neck. Functional electrical stimulation (FES) technology applies electrical current to nerves and muscles to restore movement, and controllers for upper extremity FES neuroprostheses calculate stimulation patterns to produce desired arm movement. However, currently available FES controllers have yet to restore natural movements. Reinforcement learning (RL) is a reward-driven control technique; it can employ user-generated rewards, and human preferences can be used in training. To test this concept with FES, we conducted simulation experiments using computer-generated “pseudo-human” rewards. Rewards with varying properties were used with an actor-critic RL controller for a planar two-degree-of-freedom biomechanical human arm model performing reaching movements. Results demonstrate that sparse, delayed pseudo-human rewards permit stable and effective RL controller learning. The frequency of reward is proportional to learning success, and human-scale sparse rewards permit greater learning than exclusively automated rewards. Diversity of training task sets did not affect learning. Long-term stability of trained controllers was observed. Using human-generated rewards to train RL controllers for upper-extremity FES systems may be useful. Our findings represent progress toward achieving human-machine teaming in control of upper-extremity FES systems for more natural arm movements based on human user preferences and RL algorithm learning capabilities.

Cockroach shelter-seeking strategy may look like an undirected random search, but we show that they are attracted to darkened shelters, arriving at a shelter in about half the time it would otherwise take. We were able to identify four statistically significant trends from the behavior of 134 cockroaches in one-minute naïve walking trials with four different arena configurations. By combining these trends into a model, we arrive at an algorithm that significantly directs a simulated agent to a location. This algorithm was then adapted and tested on a small mobile robot equipped with an onboard camera and antenna-like contact sensors.

Sampling-based planning algorithms of a variety of types have demonstrated pathologically poorly-performing cases, ranging from narrow passages for PRM-based roadmap methods to bug traps for RRT-based tree search methods. This paper introduces an algorithm rooted in the expansion scheme of the RRT that uses local trees to improve performance in difficult cases without sacrificing it in straightforward ones. This method interconnects these local trees, forming a roadmap that is useable for future queries. Additionally, a viable path can be trivially extracted by treating the output as a tree, or one of improved quality can be obtained via discrete search. Experimental data demonstrate performance equal to or better than several other single-query algorithms on two-dimensional test problems and significantly better on two common SE(3) benchmark problems, the flange and the alpha puzzle.

The evolution of sampling-based planning has introduced a number of novel algorithms and modifications that allow for various adaptations to the local environment. This paper presents a revision to the Path-length Annexed Random Tree (PART) that allocates and regulates local exploration in an adaptive manner. The improved algorithm eliminates the need to choose and tune an additional threshold parameter while providing denser coverage in potentially complex regions and a natural means of connection for either unidirectional or bidirectional planning. The robustness of this algorithm and the performance implications of the threshold parameter are demonstrated in 2D and SE(3) experiments.

We present heuristic algorithms for pruning large sets of candidate paths or trajectories down to smaller subsets that maintain desirable characteristics in terms of overall reachability and path length. Consider the example of a set of candidate paths in an environment that is the result of a forward search tree built over a set of actions or behaviors. The tree is precomputed and stored in memory to be used online to compute collision-free paths from the root of the tree to a particular goal node. In general, such a set of paths may be quite large, growing exponentially in the depth of the search tree. In practice, however, many of these paths may be close together and could be pruned without a loss to the overall problem of path-finding. The best such pruning for a given resulting tree size is the one that maximizes path diversity, which is quantified as the probability of the survival of paths, averaged over all possible obstacle environments. We formalize this notion and provide formulas for computing it exactly. We also present experimental results for two approximate algorithms for path set reduction that are efficient and yield desirable properties in terms of overall path diversity. The exact formulas and approximate algorithms generalize to the computation and maximization of spatio-temporal diversity for trajectories.

Architecting systems designed to elicit group-level behavior beyond the capability of any single agent, however, demands a labor and experimentation-intensive cycle on the part of the programmer. As part of a system to evolve swarm behaviors, we have developed a mixed radix fitness function to overcome the problems encountered with typical fitness functions when used in a multi-objective optimization problem. In this work, we show that mixed radix fitness functions can be used to encode sequential dependencies and prioritize metrics within the context of agent-based swarm behavior. To demonstrate the effectiveness of our approach, we construct a mixed radix fitness function and evolve swarm algorithms to solve a complex extension of the classic object collection problem. Further, we show the mixed radix fitness function is successful in driving evolution towards a feasible solution while avoiding local extrema.

The rapidly-exploring random tree (RRT) algorithm has found widespread use in the field of robot motion planning because it provides a single-shot, probabilistically complete planning method which generalizes well to a variety of problem domains. We present the multipartite RRT (MP-RRT), an RRT variant which supports planning in unknown or dynamic environments. By purposefully biasing the sampling distribution and re-using branches from previous planning iterations, MP-RRT combines the strengths of existing adaptations of RRT for dynamic motion planning. Experimental results show MP-RRT to be very effective for planning in dynamic environments with unknown moving obstacles, replanning in high-dimensional configuration spaces, and replanning for systems with space time constraints.

Actuator networks will enable an unprecedented degree of distributed control of physical environments, and further progress will critically depend on the availability of a simulation platform that can capture both the physical and the communication dynamics.

In this paper, we propose a bandwidth allocation scheme for networked control systems that have their control loops closed over a geographically distributed network. We first formulate the bandwidth allocation as a convex optimization problem. We then present an allocation scheme that solves this optimization problem in a fully distributed manner. In addition to being fully distributed, the proposed scheme is asynchronous, scalable, dynamic and flexible. We further discuss mechanisms to enhance the performance of the allocation scheme. We present analytical and simulation results

Through the use of mechanical, actuated antennae a biologically-inspired robot is capable of autonomous decision-making and navigation when faced with an obstacle that can be climbed over or tunneled under. Vertically-sweeping mechanical antennae and interface microcontrollers have been added to the Whegs ™ II [1] sensor platform that allow it to autonomously sense the presence of, and successfully navigate a horizontal shelf placed in its path. The obstacle is sensed when the antennae make contact with it, and navigation is made possible through articulation of the Whegs ™ II body flexion joint.

This paper deals with the class of robotic assemblies where position uncertainty far exceeds assembly clearance, and visual assistance is not available to resolve the uncertainty. Under this scenario, we can implement a localization strategy that resolves the uncertainty using a pre-acquired map of all possible peg-hole contact configurations. Prior to assembly, the strategy explores the contact configuration space (C-space) by sequentially bringing the peg (under different configurations) into contact with the stationary (fixtured) hole and matching the contact configurations thus recorded with the map. The different peg configurations can be actively chosen to maximize uncertainty-reduction. However, with a sampled map of the contact C-space, discretization errors are introduced, and implementing deterministic matching (at the fine-grained level necessary for assembly) would soon become prohibitively expensive in terms of computation. Additionally, with global initial uncertainty, multiple solutions abound in our localization problem. In this paper, we introduce a particle filter implementation which can not only handle the discretization errors in map-matching, but also track multiple solutions simultaneously. The particle filter implementation was validated on computer simulations of round and square peg-in-hole assemblies, before testing it on corresponding actual robotic assemblies. The implementation was highly successful on both the assemblies, reducing the uncertainty by more than 95 % and making it easy for a previously-devised compliant strategy to achieve assembly. Results from the simulations and actual assemblies are reported. We also present a comparison of these results (using random localization: peg moves selected randomly) with preliminary results from assemblies using active localization.

In this paper, we introduce several discrete space search algorithms based on continuous-space motion planning techniques such as rapidly exploring random trees (RRTs) and probabilistic roadmaps (PRMs). We describe methods for adapting these algorithms for discrete use by replacing distance metrics with cost-to-go heuristic estimates and substituting local planners for straight-line connectivity. Finally, we explore coverage and optimality properties of these algorithms in discrete spaces.

This paper presents a localization strategy for robotic assemblies with position uncertainty. The assembly of parts whose position uncertainty exceeds assembly clearance has to rely on either visual assistance or searching to achieve parts mating. We present a general strategy, applicable to arbitrary peg-in-hole assemblies, that localizes the misalignment of the mating parts in an efficient manner. The strategy explores the assembly contact configuration space and matches its observations to a pre-acquired C-space map. Simulations and experiments for various assembly scenarios are presented.

We propose the use of quasi-random sampling techniques for path planning in high-dimensional configuration spaces. Following similar trends from related numerical computation fields, we show several advantages offered by these techniques in comparison to random sampling. Our ideas are evaluated in the context of the probabilistic roadmap (PRM) framework. Two quasi-random variants of PRM- based planners are proposed: 1) a classical PRM with quasi-random sampling; and 2) a quasi-random lazy-PRM. Both have been implemented, and are shown through experiments to offer some performance advantages in comparison to their randomized counterparts.

Robotic assemblies are inherently hybrid systems. This paper pursues a class of multitiered peg-in-hole assemblies that we call "peg-in-maze" assemblies. These assemblies require a force-responsive, low-level controller governing physical contacts plus a decision-making, strategic-level supervisor monitoring the overall progress. To capture this dichotomy we formulate hybrid automata, where each state represents a different force-controlled "behavior" and transitions between states encode the high-level strategy of the assembly. Our over-arching goal is to produce a computational framework for the simulation, verification, and synthesis of such force-guided robotic assembly strategies. We investigate the use of three general hybrid-systems software tools (Hybrid cc, HyTech, and CEtool) for the simulation and verification of these strategies. We describe the computational environment we developed at Case to synthesize and implement real-world assembly strategies.

This paper presents search strategies for peg-in-hole assemblies with position uncertainty. The assembly of parts whose position uncertainty far exceeds assembly clearance has to rely on either, visual assistance or blind searching to achieve part mating. In the absence of visual assistance we look at blind search strategies to cover the search area in an efficient manner. The paper also describes a tilt strategy using special cues to guide the assembly and enhance blind search. Results from assemblies on an actual robot are presented.

We illustrate a methodology for modeling and analyzing flexible feeders using generalized semi-Markov process (GSMP) models. Working through the simple case consisting of a single part being fed on a flexible feeder, we show how the throughput of the system may be obtained by both GSMP simulation and analytical techniques for GSMP models. Further, we demonstrate the predictive capability of such models. This is accomplished by generating and validating a model of the system feeding three distinct part types (at the same time) and then modifying the model to allow other feeding scenarios to be predicted. These scenarios include the effect of feeding the parts in a specific order, the effect of using a robot with different speed capabilities, and the effect of using a different-sized presentation conveyor. We validate the predictions with physical testing.

This paper introduces the concept of a hybrid system and some of the challenges associated with the stability of such systems, including the issues of guaranteeing stability of switched stable systems and finding conditions for the existence of switched controllers for stabilizing switched unstable systems. In this endeavour, this paper surveys the major results in the (Lyapunov) stability of finite-dimensional hybrid systems and then discusses the stronger, more specialized results of switched linear (stable and unstable) systems. A section detailing how some of the results can be formulated as linear matrix inequalities is given. Stability analyses on the regulation of the angle of attack of an aircraft and on the PI control of a vehicle with an automatic transmission are given. Other examples are included to illustrate various results in this paper.

Summarizes results of a five-year, multi-disciplinary, university-industry collaborative effort investigating design issues in agile manufacturing. The focus of the project is specifically on light mechanical assembly, with the demand that new assembly tasks be implementable quickly, economically, and effectively. Key to achieving these goals is the ease of equipment and software reuse. Design choices for both hardware and software must strike a balance between the inflexibility of special-purpose designs and the impracticality of overly general designs. We review both our physical and software design choices and make recommendations for the design of agile manufacturing systems.

Flexible parts feeding techniques have begun to gain industry acceptance. However, one barrier to effective flexible feeding solutions is a dearth of knowledge of the under lying dynamics involved in flexible part feeders. The paper presents the results of testing the CWRU flexible parts feeder. Data was collected for extended periods while feeding a variety of parts. The data was examined to determine throughput and statistical properties. In addition, tests were performed to examine other aspects of the system. A new metric for specifying the throughput of vision-based flexible feeders is presented, interesting system phenomena are examined, and a statistical analysis of the data is performed.

Assembly tasks involving large position uncertainties are unsuitable for use of position-controlled robots. To automate such tasks, the assembly system must be responsive to contact forces. Issues in addressing force-responsive automated assembly include contact stability, the degree of force responsiveness required for success, the speed of a successful implementation, and the means to program a force-responsive system to perform a given assembly task. We examine these issues for robotic assembly in the context of automotive transmission components. We report on an impedance-based low-level algorithm and its interface to higher-level strategies that exhibits gentle, fast and reliable assembly of our example components.

Results obtained from experiments with task-level learning are described. The main idea of task-level learning is that a given task can be viewed as an input-output system driven by a vector of input variables or commands and responding with a vector of output variables or performance indicators. This formulation allows the application of powerful numerical methods to problems at a high-level of performance measurement: the task level. Task-level learning is studied as a paradigm than may help to program machines to learn from experience in order to: (1) perform a task better over time, (2) optimize task performance, and (3) generalize knowledge over tasks. Some extensions to the paradigm are explored. A refined model learning scheme is presented. Simulation experiments are performed to test the effects of different inverse models, different learning schemes, and different learning intervals. A framework for dealing with tasks that inherently try to minimize or maximize performance is presented.>

Mathematical properties of configuration space are presented, and algorithms invoking those properties for efficient computation of obstacles in configuration space are described. Simple elements in Cartesian space which can be transformed into configuration space rapidly are identified. Transformations of complex workspace shapes into configuration space are described in terms of multiple transformations of such simpler primitives. Computational considerations and examples are presented for the first three degrees of freedom of an industrial robot.>

Experiments in automatic collision avoidance for robots using acceleration-based reflex control are described. Acceleration-based reflex control responds to acceleration commands from higher levels, as opposed to prior reflex control approaches based on position commands from higher levels. Reflex control is exerted in configuration space and assumes either inherent dynamic decoupling of the robot links or feedback decoupling. Implementation of acceleration-based reflex control for 3-D collision avoidance using a General Electric GP132 robot is described. Collision avoidance with respect to stationary obstacles is guaranteed. In addition, performance measurements illustrate how the reflexes introduce no significant distortion to higher-level controllers when reflex action is not required.>