Goto

Collaborating Authors

Motion-Planning Chip Speeds Robots

IEEE Spectrum Robotics

If you've seen a robot manipulation demo, you've almost certainly noticed that the robot tends to spend a lot of time looking like it's not doing anything. It's tempting to say that the robot is "thinking" when this happens, and that might even be mostly correct: Odds are that you're waiting for some motion-planning algorithm to figure out how to get the robot's arm and gripper to do what it's supposed to do without running into anything. This motion-planning process is one of the most important skills a robot can have, and it's also one of the most time consuming. Researchers at Duke University, in Durham, N.C., have found a way to speed up motion planning by three orders of magnitude while using one-twentieth the power. Their solution is a custom processor that can perform the most time-consuming part of the job--checking for all potential collisions across the robot's entire range of motion--with unprecedented efficiency.


PA*SE: Parallel A* for Slow Expansions

AAAI Conferences

Planners need to become faster as we seek to tackle increasingly complicated problems. Much of the recent improvements in computer speed is due to multi-core processors. For planners to take advantage of these types of architectures, we must adapt algorithms for parallel processing. There are a number of planning domains where state expansions are slow. One example is robot motion planning, where most of the time is devoted to collision checking. In this work, we present PA*SE, a novel, parallel version of A* (and weighted A*) which parallelizes state expansions by taking advantage of this property. While getting close to a linear speedup in the number of cores, we still preserve completeness and optimality of A* (bounded sub-optimality of weighted A*). PA*SE applies to any planning problem in which significant time is spent on generating successor states and computing transition costs. We present experimental results on a robot navigation domain (x,y,heading) which requires expensive 3D collision checking for the PR2 robot. We also provide an in-depth analysis of the algorithm’s performance on a 2D navigation problem as we vary the number of cores (up to 32) as well as the time it takes to collision check successors during state expansions.


Minimizing Conflicts Between Moving Agents over a Set of Non-Homotopic Paths Through Regret Minimization

AAAI Conferences

This paper considers a game-theoretic framework for motion coordination challenges. The focus of this work is to minimize the number of interactions agents have when moving through an environment. In particular, agents employ a replanning framework and regret minimization over a set of actions, which correspond to different homotopic paths. By associating a cost to each trajectory, a motion coordination game arises. Regret minimization is argued as an appropriate technique, as agents do not have any knowledge of other agents' cost functions. This work outlines a methodology for minimizing the regret of actions in a computationally efficient way. Initial simulation results involving pairs of mobile agents show indications that the proposed framework can improve the choice of non-colliding paths compared to a greedy choice by the agents, without increasing any information requirements.


g-Planner: Real-time Motion Planning and Global Navigation using GPUs

AAAI Conferences

We present novel randomized algorithms for solving global motion planning problems that exploit the computational capabilities of many-core GPUs. Our approach uses thread and data parallelism to achieve high performance for all components of sample-based algorithms, including random sampling, nearest neighbor computation, local planning, collision queries and graph search. The approach can efficiently solve both the multi-query and single-query versions of the problem and obtain considerable speedups over prior CPU-based algorithms. We demonstrate the efficiency of our algorithms by applying them to a number of 6DOF planning benchmarks in 3D environments. Overall, this is the first algorithm that can perform real-time motion planning and global navigation using commodity hardware.


Cooperative, Dynamics-based, and Abstraction-Guided Multi-robot Motion Planning

Journal of Artificial Intelligence Research

This paper presents an effective, cooperative, and probabilistically-complete multi-robot motion planner that enables each robot to move to a desired location while avoiding collisions with obstacles and other robots. The approach takes into account not only the geometric constraints arising from collision avoidance, but also the differential constraints imposed by the motion dynamics of each robot. This makes it possible to generate collision-free and dynamically-feasible trajectories that can be executed in the physical world.The salient aspect of the approach is the coupling of sampling-based motion planning to handle the complexity arising from the obstacles and robot dynamics with multi-agent search to find solutions over a suitable discrete abstraction. The discrete abstraction is obtained by constructing roadmaps to solve a relaxed problem that accounts for the obstacles but not the dynamics. Sampling-based motion planning expands a motion tree in the composite state space of all the robots by adding collision-free and dynamically-feasible trajectories as branches. Efficiency is obtained by using multi-agent search to find non-conflicting routes over the discrete abstraction which serve as heuristics to guide the motion-tree expansion. When little or no progress is made, the routes are penalized and the multi-agent search is invoked again to find alternative routes. This synergistic coupling makes it possible to effectively plan collision-free and dynamically-feasible motions that enable each robot to reach its goal. Experiments using vehicle models with nonlinear dynamics operating in complex environments, where cooperation among robots is required, show significant speedups over related work.