Learning to guide task and motion planning using score-space representation

As robotic systems evolve and get more sophisticated, expectations of them to accomplish high-level tasks increase gradually and their motion planning becomes more complex and difficult. The motion planning problem has been studied for more than four decades from different aspects such that presently has a vast literature. This paper investigates different components of the robot motion planning (RMP) problem and presents a new comprehensive taxonomy for a wide range of RMP problems. The taxonomy is based on a survey of the literature on RMP problems and applications in robotics and computer science.

In this work we present a methodology for intelligent motion planning in an uncertain environment using a non-local sensor, such as radar. This methodology is applied to an unmanned helicopter navigating a cluttered urban environment. We show that the problem of motion planning in a uncertain environment, under certain assumptions, can be posed as the adaptive optimal control of an uncertain Markov decision process, characterized by a known, control dependent system, and an unknown, control independent environment. The strategy for motion planning then reduces to computing the control policy based on the current estimate of the environment, also known as the "certainty equivalence principle" in the adaptive control literature. Our methodology allows the inclusion of a non-local sensor into the problem formulation, which significantly accelerates the convergence of the estimation and planning algorithms. Further we show that the motion planning and estimation problems, as formulated in this paper possess special structure which can be exploited to significantly reduce the computational burden of the associated algorithms. We apply this methodology to the problem of motion planning for an unmanned helicopter in a partially known model of the Texas A&M campus

A methodology is presented in this work for intelligent motion planning in an uncertain environment using a nonlocal sensor, such as a radar. This methodology is applied to an unmanned helicopter navigating a cluttered urban environment. It is shown that the problem of motion planning in an uncertain environment, under certain assumptions, can be posed as the adaptive optimal control of an uncertain Markov decision process, characterized by a known, control dependent system, and an unknown, control independent environment. The strategy for motion planning then reduces to computing the control policy based on the current estimate of the environment, also known as the certainty equivalence principle in the adaptive control literature. The methodology allows the inclusion of a nonlocal sensor into the problem formulation, which significantly accelerates the convergence of the estimation and planning algorithms. Further, the motion planning and estimation problems possess special structure which can be exploited to significantly reduce the computational burden of the associated algorithms. The methodology is applied to the problem of motion planning for an unmanned helicopter through a partially known model of the Texas A&M campus and the testing is done on a flight simulator.

As for fishlike robot propulsion,there exist a number of possible motion modes to achieve a prescribed swimming velocity. But for optimum efficiency,there is only one mode. The motion planning on optimum efficiency of fishlike robot is presented to achieve a prescribed velocity. The steady state swimming can be realized and the intrinsic rules for the steady velocity,power consumption and propulsive efficiency with respect to different motion modes are revealed by using the numerical simulation method. The problem of motion planning is formulated to find the motion mode that maximizes the efficiency over the motion space,which is defined by oscillating frequency f and maximum amplitude Amax,subject to the prescribed velocity. The results show that the motion mode with lower f and higher Amax can obtain optimum efficiency while subjected to the prescribed velocity. Taking 2 m·s–1 for example,the value of efficiency for the motion mode with lower f and higher Amax increases by 37.7% compared with that of higher f and lower Amax,and this trend is clearly reflected by the three-dimensional flow structure. The findings are of great significance to the acquirement of efficient motion laws,understanding of swimming mechanism and the propulsive motion design for fishlike robot.

This study is devoted to the challenges of motion planning for mobile robots with smart machine vision systems. Motion planning for mobile robots in the environment with obstacles is a problem to deal with when creating robots suitable for operation in real-world conditions. The solutions found today are predominantly private, and are highly specialized, which prevents judging of how successful they are in solving the problem of effective motion planning. The purpose of this study: develop and investigate a motion planning algorithm for a mobile robot with a smart machine vision system. The research subject for this article is a motion planning algorithm for a mobile robot with a smart machine vision system. This study provides a review of domestic and foreign mobile robots that solve the motion planning problem in a known environment with unknown obstacles. The following navigation methods are considered for mobile robots: local, global, individual. In the course of work and research, a mobile robot prototype has been built, capable of recognizing obstacles of regular geometric shapes, as well as plan and correct the movement path. Environment objects are identified and classified as obstacles by means of digital image processing methods and algorithms. Distance to the obstacle and relative angle are calculated by photogrammetry methods, image quality is improved by linear contrast enhancement and optimal linear filtering using the Wiener-Hopf equation. Virtual tools, related to mobile robot motion algorithm testing, have been reviewed, which led us to selecting Webots software package for prototype testing. Conclusions have been drawn regarding the concluded research.

This study is devoted to the challenges of motion planning for mobile robots with smart machine vision systems. Motion planning for mobile robots in the environment with obstacles is a problem to deal with when creating robots suitable for operation in real-world conditions. The solutions found today are predominantly private, and are highly specialized, which prevents judging of how successful they are in solving the problem of effective motion planning. Solutions with a narrow application field already exist and are being already developed for a long time, however, no major breakthrough has been observed yet. Only a systematic improvement in the characteristics of such systems can be noted. The purpose of this study: develop and investigate a motion planning algorithm for a mobile robot with a smart machine vision system. The research subject for this article is a motion planning algorithm for a mobile robot with a smart machine vision system. This study provides a review of domestic and foreign mobile robots that solve the motion planning problem in a known environment with unknown obstacles. The following navigation methods are considered for mobile robots: local, global, individual. In the course of work and research, a mobile robot prototype has been built, capable of recognizing obstacles of regular geometric shapes, as well as plan and correct the movement path. Environment objects are identified and classified as obstacles by means of digital image processing methods and algorithms. Distance to the obstacle and relative angle are calculated by photogrammetry methods, image quality is improved by linear contrast enhancement and optimal linear filtering using the Wiener-Hopf equation. Virtual tools, related to mobile robot motion algorithm testing, have been reviewed, which led us to selecting Webots software package for prototype testing. Testing results allowed us to make the following conclusions. The mobile robot has successfully identified the obstacle, planned a path in accordance with the obstacle avoidance algorithm, and continued moving to the destination. Conclusions have been drawn regarding the concluded research.

A problem of motion planning and coordination of compact formations of ground and aerial robots will be tackled in this paper. The scenarios when the formation composed from Unmanned Ground Vehicles (UGVs) and Unmanned Aerial Vehicles (UAVs), in particular Micro Aerial Vehicles (MAVs), has to reverse the direction of movement to fulfil task of collisionfree motion to a target zone will be solved. The presented motion planning and stabilization approach provides an effective technique to enable deployment of closely cooperating teams of robots in outdoor as well as indoor environment. The formation to target region problem is solved using a Model Predictive Control (MPC) methodology and the formation driving concept is based on a virtual-leader-follower approach. The mentioned MPC based process is used for trajectory planning and control of a virtual leader and also for control and stabilization of followers (MAVs and UGVs). The proposed approach is verified with numerous simulations and hardware experiments.

Bi-objective path planning using deterministic algorithms

This paper presents an approach to coordinating a group of mobile robots that move along the designed paths. Subject to the constraints of velocity/acceleration bounds and collision avoidance, the robots are required to maintain the formation relationship while moving. The coordination is realized by posing the coordination problem as a velocity optimization problem in motion planning and then properly planning robots' velocities along the paths. The linear interactive and general optimizer (lingo) is utilized to achieve the velocity optimization goal. Simulations are performed on a group of mobile robots to demonstrate the effectiveness of the proposed approach in coordinated motion planning with the formation requirement.

This paper addresses a problem of optimal motion planning of mobile platforms amidst obstacles, considering the mobile platform dynamics. Due to nonholonomic constraints, actuator constraints, and state constraints by obstacle avoidance, the planning problem of mobile platform with two independently driven wheels is a complicated one. In this study, a dynamical model for the mobile platform is presented, including nonholonomic kinematic constraints. The idea of a path parameter is introduced to simplify the planning problem by considering the dynamics and nonholonomic constraints. Using the path parameter, the optimal motion planning problem is divided into two sub-problems: 1) time-optimization of trajectory along specified path, and 2) search for optimal path. Then two methods are proposed the solve the problems using the path parameter and parametrization by B-spline function. Finally, quasi-time-optimal solution for the original problem are planned by combining the two methods. Numerical examples show effectiveness of the motion planner.

The potential method is superior for solving the problem of motion planning; however, it must address the problem of the real-time generation of potential field. Obstacle avoidance is a motion planning problem. In a previous study, we investigated the real-time generation of potential field. Based on parallel processing with element group, we proposed the system by Sensory Point Moving (SPM) method. As a result of computer simulation, it was confirmed that the SPM method is effective for generating an obstacle avoidance path in 2-D and a more complex working environment like a 3-D one. In this paper, we discuss the development of autonomous mobile robot for obstacle avoidance based on the SPM method.

Trajectory planning in dynamic environments can be decomposed into two sub-problems: 1) planning a path to avoid static obstacles, 2) then planning a speed profile to avoid dynamic obstacles. This is also called path-speed decomposition. In this work, we present a novel approach to solve the first sub-problem, motion planning with static obstacles. From an optimization perspective, motion planning for autonomous vehicles can be viewed as non-convex constrained nonlinear optimization, which requires a good enough initial guess to start and is often sensitive to algorithm parameters. We formulate motion planning as convex spline optimization. The convexity of the formulated problem makes it able to be solved fast and reliably, while guaranteeing a global optimum. We then reorganize the constrained spline optimization into a recurrent formulation, which further reduces the computational time to be linear in the optimization horizon size. The proposed method can be applied to both trajectory generation and motion planning problems. Its effectiveness is demonstrated in challenging scenarios such as tight lane changes and sharp turns.

SUMMARYThis paper presents an architecture for the navigation of an autonomous mobile robot evolving in environments with obstacles. Instead of addressing the motion planning and control problems in different contexts, these issues are described in connected modules with performance requirement considerations. The path planning problem is formulated as a constrained receding horizon planning problem and is solved in real time with an efficient computational method that combines non-linear control theory, B-spline basis function and non-linear programming. An integral sliding mode controller is used for trajectory tracking. The closed-loop stability of the tracking errors is guaranteed in spite of unknown disturbances. It is also shown that this strategy is particularly useful if integral sliding mode control is combined with other methods to further robustify against perturbations. The effectiveness, perfect performance of obstacle avoidance, real time and high robustness properties are demonstrated by experimental results.

Robots operating in the real world must combine task planning for reasoning about what to do with motion planning for reasoning about how to do it -- this is known as task and motion planning. One promising approach for task and motion planning is Logic Geometric Programming (LGP) which integrates a logical layer and a geometric layer in an optimization formulation. The logical layer describes feasible high-level actions at an abstract symbolic level, while the geometric layer uses continuous optimization methods to reason about motion trajectories with geometric constraints. In this paper we propose a new approach for solving task and motion planning problems in the LGP formulation, that leverages state-of-the-art diverse planning at the logical layer to explore the space of feasible logical plans, and minimizes the number of optimization problems to be solved on the continuous geometric layer. To this end, geometric infeasibility is fed back into planning by identifying prefix conflicts and incorporating this back into the planner through a novel multi-prefix forbidding compilation. We further leverage diverse planning with a new novelty criteria for selecting candidate plans based on the prefix novelty, and a metareasoning approach which attempts to extract only useful conflicts by leveraging the information that is gathered in the course of solving the given problem.

We propose Dynamic-goal Deep Reinforcement Learning (DGDRL) method to address the problem of robot arm motion planning in telemanipulation applications. This method intuitively maps human hand motions to a robot arm in real-time, while avoiding collisions, joint limits and singularities. We further propose a novel hardware setup, based on the HTC VIVE VR system, that enables users to smoothly control the robot tool position and orientation with hand motions, while monitoring its movements in a 3D virtual reality environment. A VIVE controller captures 6D hand movements and gives them as reference trajectories to a deep neural policy network for controlling the robot’s joint movements. Our DGDRL method leverages the state-of-art Proximal Policy Optimization (PPO) algorithm for deep reinforcement learning to train the policy network with the robot joint values and reference trajectory observed at each iteration. Since training the network on a real robot is time-consuming and unsafe, we developed a simulation environment called RobotPath which provides kinematic modeling, collision analysis and a 3D VR graphical simulation of industrial robots. The deep neural network trained using RobotPath is then deployed on a physical robot (ABB IRB 120) to evaluate its performance. We show that the policies trained in the simulation environment can be successfully used for trajectory planning on a real robot. The the codes, data and video presenting our experiments are available at https://github.com/kavehkamali/ppoRobotPath.