SUMMARY This work presents an approach to motion planning for robotic manipulators that aims at improving path quality in terms of safety. Safety is explicitly assessed using the quantity called danger field. The measure of safety can easily be embedded into a heuristic function that guides the exploration of the free configuration space. As a result, the resulting path is likely to have substantially higher safety margin when compared to one obtained by regular planning algorithms. To this end, four planning algorithms have been proposed. The first planner is based on volume trees comprised of bubbles of free configuration space, while the remaining ones represent modifications of classical sampling-based algorithms. Several numerical case studies are carried out to validate and compare the performance of the presented algorithms with respect to classical planners. The results indicate significantly lower danger metric for paths obtained by safety-oriented planners even with some decrease in running time.

This paper presents a new control law for robotic manipulators in unstructured environments which guarantees the achievement of the goal position without incurring in local minima. The passivity of the closed-loop system renders this control scheme well-suited for human-robot coexistence, especially when the robot is supposed to share its workspace with humans. The given control law has been implemented and experimentally tested in a realistic scenario, demonstrating the effectiveness in driving the robot to a given configuration in a cluttered environment without any offline planning phase.

This paper presents a strategy for sensor-based trajectory generation in unstructured environments which guarantees the achievement of the goal position without incurring in local minima. The passivity of the closed-loop system renders this control scheme well-suited for human-robot cooperation, especially when the robot is supposed physically interact with humans. The given control law has been implemented and experimentally tested in a realistic scenario, demonstrating the effectiveness in driving the robot to a given configuration in a cluttered environment without any offline planning phase.

We present a sampling-based motion planning approach for articulated manipulators that generates safe paths. It uses the rapidly-exploring random trees paradigm to establish a collision-free path in configuration space. The expansion of the trees is influenced by a modified version of the kinetostatic danger field — a safety assessment function recently proposed in the literature. The idea is to grow the trees towards safer regions. Thus, the planner provides not only collision-free paths, but strives for safer ones. We propose two versions of the planner. The first is a modification of the Jacobian Transpose-directed RRT (JT-RRT) algorithm that grows a single tree from the start configuration and uses the transpose of the Jacobian to guide the sampling towards the goal defined in the workspace. The second is an extension of the standard bidirectional RRT-connect planner where the inputs to the algorithm are the start and the goal configuration that serve as seeds for the trees growth.

This paper deals with controller design for gentle physical human-robot interaction. Two objectives are set up. The first is to establish an analytical framework in order to justify the good features of state of the art controller, recently designed by numerical search of parameter space. The second is to investigate the possibilities to improve the performance of such controller. Our method ensures “prescribed” admittance behavior of the robot, similar to natural admittance controller design but with both more realistic model of the robot and more realistic target admittance. Joining natural admittance approach with the concept of complementary stability allows reaping the benefits of both. Limited knowledge about the environment via structured uncertainty allows a very simple worst-case analysis using elementary tools such as Routh–Hurwitz stability criterion. Consequent relation within the parameters determines an allowed region in the parameter space, where the contact stability is guaranteed. Not surprisingly, on one border of this region, the system behaves exactly the same as when the state of the art controller is employed. In addition, unexpected stability regions are discovered, suggesting theoretical performance improvements.

Abstract In this paper, a control strategy for improving safety in human-robot interaction is described. The approach is based on the concept of kinetostatic danger field—a safety assessment recently proposed in the literature. A method for mapping the danger field information directly into position/velocity commands, thus bypassing the dynamics of the manipulator, is presented. Such an approach is suitable for industrial manipulators that usually require decentralized position control. Moreover, decoupling of task and posture behavior enables safety enhancement without compromising the task. The proposed control strategy is validated within a simulation study.

This paper presents a framework for planning and control for the safe behavior of the robotic manipulators. The framework exploits the relations between robot, human and the environment to measure safety via quantity called danger field. According to the degree of danger, the robot reacts promptly, improving the human safety. Above the danger field based control, there is a global path planner that does not only provide the collision-free paths but strives for safer ones.

This paper presents a novel method for evaluating the danger within the environment of a robot manipulator. It is based on the introduced concept of kinetostatic danger field, a quantity that captures the complete state of the robot - its configuration and velocity. The field itself is invariant with respect to objects around the robot and can be computed in any given point of the workspace using measurements from the proprioceptive sensors. Moreover, all the computation can be performed in closed form, yielding compact algebraic expressions that allow for real time applications. The danger field is not only a meaningful indicator about the risk in the vicinity of the robot, but can also be fed back within control skills that implement some well known safety strategies like collision avoidance and virtual impedance control, provided that some environment perception is available in order to determine the points where the field should be computed. Kinematic redundancy for simultaneous task performance and danger minimization can be exploited. The methodology described in the paper is supported with simulation results.

We propose a novel method of path planning for robotic manipulators that is based on the tree expansion via bubbles of free configuration space. The algorithm is designed to yield collision-free paths that also tend to minimize a certain danger criterion. This is achieved by embedding a suitably tailored heuristics within the algorithm. For that purpose we use a recently proposed safety assessment based on the concept of the danger field - an easily computable quantity that captures the complete kinematic behavior of the manipulator. Under the assumption that a systematic graph search technique dictates the tree growth, we prove the algorithm's completeness.

A novel method to obtain safe paths for robotic manipulators is presented. It emulates the probabilistic roadmap planner by trying to establish a collision-free path that connects the start and the goal configuration via samples that belong to a quasi-random, low discrepancy sequence. The path construction is guided by a search algorithm with a heuristic function that includes a suitably tailored safety estimation. The measure of safety is based on the danger field - a recently proposed safety assessment for human-robot interaction. Thus, the planner provides not only collision-free paths, but strives for safer ones. In order to decrease the expected number of collision checks, we propose a novel method for testing whether local paths are collision-free or not. The method combines the standard binary collision checking with the concept of bubbles of free configuration space.

In this paper we define a mathematical notion of ectropy for classifying diversity measures in terms of the extent to which they tend to penalize point collocation, we investigate the advantages and disadvantages of several known measures and we propose some novel ones. In particular, we introduce a measure based on Euclidean minimum spanning trees, a class of power mean based measures and three measures based on discrepancy from uniform distribution. All considered measures are tested and compared on a large set of random and structured populations. Special attention is also devoted to the complexity of computing the measures. The measure based on Euclidean minimum spanning trees turns out to be the most promising one in terms of the tradeoff between the computational complexity and the ectropic behavior.

In a modern vehicle systems one of the main goals to achieve is driver's safety, and many sophisticated systems are made for that purpose. Vibration isolation for the vehicle seats, and at the same time for the driver, is one of the challenging problems. Parameters of the controller used for the isolation can be tuned for a different road types, making the isolation better (specially for the vehicles like dampers, tractors, field machinery, bulldozers, etc.). In this paper we propose the method where neural networks are used for road type recognition. The main goal is to obtain a good road recognition for the purpose of better vibration damping of a driver's semi active controllable seat. The recognition of a specific road type will be based on the measurable parameters of a vehicle. Discrete Fourier Transform of measurable parameters is obtained and used for the neural network learning. The dimension of the input vector, as the main parameter that decides the speed of road recognition, is varied.