Planning a safe trajectory for an ego vehicle through an environment with occluded regions is a challenging task. Existing methods use some combination of metrics to evaluate a trajectory, either taking a worst case view or allowing for some probabilistic estimate, to eliminate or minimize the risk of collision respectively. Typically, these approaches assume occluded regions of the environment are unsafe and must be avoided, resulting in overly conservative trajectories-particularly when there are no hidden risks present. We propose a local trajectory planning algorithm which generates safe trajectories that maximize observations on un-certain regions. In particular, we seek to gain information on occluded areas that are most likely to pose a risk to the ego vehicle on its future path. Calculating the information gain is a computationally complex problem; our method approximates the maximum information gain and results in vehicle motion that remains safe but is less conservative than state-of-the-art approaches. We evaluate the performance of the proposed method within the CARLA simulator in different scenarios.

We propose a new scheme to learn motion planning constraints from human driving trajectories. Behavioral and motion planning are the key components in an autonomous driving system. The behavioral planning is responsible for high-level decision making required to follow traffic rules and interact with other road participants. The motion planner role is to generate feasible, safe trajectories for a self-driving vehicle to follow. The trajectories are generated through an optimization scheme to optimize a cost function based on metrics related to smoothness, movability, and comfort, and subject to a set of constraints derived from the planned behavior, safety considerations, and feasibility. A common practice is to manually design the cost function and constraints. Recent work has investigated learning the cost function from human driving demonstrations. While effective, the practical application of such approaches is still questionable in autonomous driving. In contrast, this paper focuses on learning driving constraints, which can be used as an add-on module to existing autonomous driving solutions. To learn the constraint, the planning problem is formulated as a constrained Markov Decision Process, whose elements are assumed to be known except the constraints. The constraints are then learned by learning the distribution of expert trajectories and estimating the probability of optimal trajectories belonging to the learned distribution. The proposed scheme is evaluated using NGSIM dataset, yielding less than 1% collision rate and out of road maneuvers when the learned constraints is used in an optimization-based motion planner.

Motion planning is one of the key modules in autonomous driving systems to generate trajectories for self-driving vehicles to follow. A common motion planning approach is to generate trajectories within semantic safe corridors. The trajectories are generated by optimizing parametric curves (e.g. Bezier curves) according to an objective function. To guarantee safety, the curves are required to satisfy the convex hull property, and be contained within the safety corridors. The convex hull property however does not necessary hold for time-dependent corridors, and depends on the shape of corridors. The existing approaches only support simple shape corridors, which is restrictive in real-world, complex scenarios. In this paper, we provide a sufficient condition for general convex, spatio-temporal corridors with theoretical proof of guaranteed convex hull property. The theorem allows for using more complicated shapes to generate spatio-temporal corridors and minimizing the uncovered search space to $O\left(\frac{1}{n^{2}}\right)$ compared to O(1) of trapezoidal corridors, which can improve the optimality of the solution. Simulation results show that using general convex corridors yields less harsh brakes, hence improving the overall smoothness of the resulting trajectories.

Motion Planning is one of the key modules in autonomous driving systems to generate trajectories for self-driving vehicles. Spatio-temporal motion planners are often used to tackle complicated and dynamic driving scenarios. While effective in dealing with temporal changes in the environment, the existing methods are limited to optimizing a particular family of cost functions defined based on decoupled longitudinal and lateral terms. However, the planning objectives can only be explained using coupled terms in some cases, e.g. closeness to the reference path, lateral acceleration, and heading rate. The limitation arises from expressing such objectives as linear and quadratic terms suitable for optimization. This paper proposes an approach with theoretical proofs to approximate the upper bound of a given couple, nonlinear cost term with a set of uncoupled terms, allowing for converting the planning optimization problem into a linear quadratic optimization. The effectiveness of the proposed approach is shown through a series of simulated scenarios. The proposed approach results in smoother and steadier trajectories in the spatial plane.

This paper presents a novel method to generate spatial constraints for motion planning in dynamic environments. Motion planning methods for autonomous driving and mobile robots typically need to rely on the spatial constraints imposed by a map-based global planner to generate a collision-free trajectory. These methods may fail without an offline map or where the map is invalid due to dynamic changes in the environment such as road obstruction, construction, and traffic congestion. To address this problem, triangulation-based methods can be used to obtain a spatial constraint. However, the existing methods fall short when dealing with dynamic environments and may lead the motion planner to an unrecoverable state. In this paper, we propose a new method to generate a sequence of channels across different triangulation mesh topologies to serve as the spatial constraints. This can be applied to motion planning of autonomous vehicles or robots in cluttered, unstructured environments. The proposed method is evaluated and compared with other triangulation-based methods in synthetic and complex scenarios collected from a real-world autonomous driving dataset. We have shown that the proposed method results in a more stable, long-term plan with a higher task completion rate, faster arrival time, a higher rate of successful plans, and fewer collisions compared to existing methods.

Motion planning under uncertainty is one of the main challenges in developing autonomous driving vehicles. In this work, we focus on the uncertainty in sensing and perception, resulted from a limited field of view, occlusions, and sensing range. This problem is often tackled by considering hypothetical hidden objects in occluded areas or beyond the sensing range to guarantee passive safety. However, this may result in conservative planning and expensive computation, particularly when numerous hypothetical objects need to be considered. We propose a reinforcement learning (RL) based solution to manage uncertainty by optimizing for the worst case outcome. This approach is in contrast to traditional RL, where the agents try to maximize the average expected reward. The proposed approach is built on top of the Distributional RL with its policy optimization maximizing the stochastic outcomes’ lower bound. This modification can be applied to a range of RL algorithms. As a proof-of-concept, the approach is applied to two different RL algorithms, Soft Actor-Critic and DQN. The approach is evaluated against two challenging scenarios of pedestrians crossing with occlusion and curved roads with a limited field of view. The algorithm is trained and evaluated using the SUMO traffic simulator. The proposed approach yields much better motion planning behavior compared to conventional RL algorithms and behaves comparably to humans driving style.

Tendon-driven mechanisms used in robotic surgery exhibit strong nonlinearities, particularly a static backlash-like hysteresis, in their motion transmission behavior. In this paper, an extension of a previously developed model is proposed that allows for estimation of angular displacements in multi-DOF tendon-driven devices where special attention is given to the coupling effect between DOFs. The proposed model consists of the conventional coupling matrix and a novel elongation matrix which compensates for the coupled hysteretic effect. The model is applied to the problem of position estimation in three DOFs (one pitch and two grasping DOFs) of a da Vinci®surgical instrument. As a further extension, a preliminary dynamic model is also suggested to deal with high-frequency inputs. According to the experimental results obtained, the proposed quasi-static model can describe the transmission behavior with goodness-of-fit of 76-92 per cent, and the estimates are improved by 35-72 per cent in terms of the RMSE for the proposed dynamic model as compared to the conventional rigid model.

In recent years, Magneto-Rheological (MR) clutches have been increasingly used for realizing compliant actuation. One difficulty in using MR clutches is the existence of nonlinear hysteretic behaviors between the input current and output torque of an MR clutch. In this paper, a new closed-loop, Field-Programable-Gate-Array (FPGA) based control scheme to linearize an MR clutch's input-output relationship is presented. The feedback signal used in this control scheme is the magnetic field acquired from hall sensors within the MR clutch. The FPGA board uses this feedback signal to compensate for the nonlinear behavior of the MR clutch using an estimated model of the clutch magnetic field. The local use of an FPGA board will dramatically simplify the use of MR clutches for torque actuation. The effectiveness of the proposed technique is validated using an experimental platform that includes an MR clutch as part of a compliant actuation mechanism. The results clearly demonstrate that the use of the FPGA based closed-loop control scheme can effectively eliminate hysteretic behaviors of the MR clutch, allowing to have linear actuators with predictable behaviors.

In this paper, an optimization method for designing MR clutches is studied. The proposed method optimizes the geometrical dimensions of an MR clutch, hence its mass, for given output torque and electrical input power. The main idea behind this optimization is that the input power and output torque are two parameters that are normally known to the designer prior to the design of an MR clutch and considering these parameters in the optimization as fixed values has a practical significance. Having presented the optimization method, we compare the characteristics of three different MR clutch configurations in order to demonstrate the effectiveness of the proposed method. A comparison between the drum, single-disk and multi-disk configurations of MR clutches is performed. Using the proposed method one can select a suitable configuration as well as the geometrical dimensions for an MR clutch that best suits the requirements of each individual design.

The two main objectives in designing a haptic interface are stability and transparency. The dynamics of the actuators employed in a haptic interface have a significant effect on these goals. In this article, the potential benefits of Magneto-Rheological Fluid (MRF) based actuators to the field of haptics are discussed. Devices developed with such fluids are known to possess superior mechanical characteristics over conventional servo systems. This contributes significantly to improved stability and transparency of haptic devices. In this study, this idea is evaluated from both theoretical and experimental points of view. First, the properties of such actuators which motivated this research are discussed. Next, two single degrees-of-freedom (DOF) haptic interfaces are used in a virtual wall experiment. These devices take advantage of an MRF-based clutch and a brushless DC motor at their core, respectively. The results of both devices are compared and show the superiority of the MRF-based clutch. In addition, design and analysis of a small-scale MRF-based clutch, suitable for a multi-DOF haptic interface, is given and its torque capacity, inertia, and mass are compared with those of conventional servo systems. Conclusions drawn from this investigation indicate that MRF clutch actuation approaches can indeed be developed to design haptic interfaces with improved stability and transparency.

In this paper, the presence of limit cycles in the behavior of antagonistically coupled Magneto-Rheological (MR) actuators is investigated. The actuator considered in this paper was developed and described in [1] and [2]. This actuator offers high torque-to-mass and torque-to-inertia ratios, for inherent safe actuation. While the antagonistic arrangement is beneficial in improving the actuator performance and eliminating backlash, it may result in limit cycles when the actuator operates in a position control loop. The occurrence of limit cycle depends on the parameters of the actuator as well as the controller. An in-depth analysis is carried out in this paper to establish a connection between the system parameters and the limit cycle occurrence. Moreover, sufficient conditions for avoiding limit cycle are derived specifically for a Proportional-Derivative (PD) controller. Simulations and experimental results validate the analysis and provide insights into the limit cycle observed in the operation of antagonistic MR actuators.

This paper presents the design and development of a novel two Degrees-Of-Freedom (DOF) safe robot manipulator. Magneto-Rheological (MR) clutches are incorporated in the design to enable antagonistic actuation at the joints. A single unidirectional motor supports bidirectional actuation of all joints. Unlike most current safe robots, high quality actuation is preserved, while the manipulator weight and effective inertia are reduced. This is achieved by relocating the driving motor to the base of the robot. Moreover, MR clutches have been shown to pose superior torque to mass, and torque to inertia characteristics over conventional servo motors, further contributing to the reduction of manipulator mass and inertia. The manipulator exhibits both high performance and intrinsic safety as a result of mechanically passive dynamics. A set of experiments is performed to validate the manipulator performance.

In this paper, adaptive compensation of the hysteresis in a Magneto-Rheological (MR) fluid based actuators and its application for sensor-less high fidelity force/torque control is investigated. The MR actuator considered in this paper was originally described in [1] and [2]. This actuator offers high torque-to-mass and torque-to-inertia ratios. Yet, as an essential component of MR actuators, the magnetic circuit of the actuator shows hysteresis between its input current/voltage and output magnetic field. The hysteresis in the magnetic circuit results in a similar relationship between the input current and the output torque of the MR actuator. The control scheme used with actuators possessing hysteresis often requires compensating for the hysteresis. To this end, we propose an adaptive control method based on feedback linearization that estimates both hysteresis and uncertain parameters of the magnetic circuit. A set of experiments is performed to validate the effectiveness of the proposed method.

In this paper, a new open-loop model for a Magneto-Rheological (MR) based actuator is presented. The model consists of two parts relating the output torque of the actuator to its internal magnetic field, and the internal magnetic field to the applied current. Each part possesses its own hysteretic behavior. The first part uses an open-loop Bouc-Wen model to relate the output torque to internal magnetic field. The second part uses a novel nonlinear adaptive observer that relates the internal magnetic field to the applied current. The model facilitates accurate control of the actuator using its input current. It also eliminates the need for force/torque sensors for providing feedback signals. The accuracy of the constructed model is validated through simulations. The overall model as well as each part of it is assessed against a widely accepted hysteresis modeling approach, known as the Preisach model and its advantages are highlighted. The second part of model is also compared to Bingham model which has been broadly employed in modeling of MR fluid dynamic. Bouc-Wen model shows higher accuracy in capturing hysteretic behavior of MR fluid in comparison to non-hysteretic Bingham model. Experimental results using the prototyped actuation mechanism further verify the accuracy of the model and demonstrate its effectiveness.

In this paper, the observer-based stabilization of unstable equilibrium points of a class of unknown nonlinear systems is proposed. The controller is based on feedback linearization where the observer system and control signal are directly estimated by a nonlinear in parameter neural network (NLPNN). A modified back propagation (BP) algorithm with e-modification was used to update the weights of the network. Globally uniformly ultimately boundedness of overall closed-loop system is ensured using Lyapunov's direct method. To verify the effectiveness of the proposed observer-based controller, a set of simulations was performed on a Rossler and Lorenz chaotic systems.

The observer-based tracking control problem for a class of nonlinear affine systems using neural networks is proposed in this paper. The controller is based on feedback linearization where the observer system and control signal are directly estimated by a nonlinear in parameter neural networks (NLPNN). A Hebbian-like algorithm with e-modification is used to update the weights of the network. The uniformly ultimately boundedness of the tracking error and all signals in the overall closed-loop system is proved using Lyapunov's direct method. To evaluate the performance of the proposed observer-based controller, a set of simulations is performed on a nonlinear cart-pole system. Simulation results show the effectiveness of the proposed control methodology.

This paper proposes a local optima avoidable particle swarm optimization (LOAPSO) which remarkably outperforms the standard PSO in the sense that it can avoid entrapment in local optimum. Three benchmark functions are used to validate the proposed algorithm and compare its performance with that of the other algorithms known as hybrid PSOs and six functions reported in SIS2005 are used to better verification of the proposed algorithm. Numerical results indicate that LOAPSO is considerably competitive due to its ability to avoid being trapped in local optima and to find the functions' global optimum as well as better convergence performance.