Safe real-time path planning for fast four-legged robots
Continuous-Time Safe Planning and Control for Agile Quadrupeds
Robotics
Summary
Fast four-legged robots need safe paths to move through changing spaces without bumping into things. The authors made a new way for these robots to plan their steps continuously and safely, even with complex movements that older simpler methods ignored. They use a type of mathematical curve to make sure the robot never hits anything at any point in its path. Their method runs quickly enough to adjust as the robot moves and has been tested in fake environments and real robots avoiding moving obstacles.
What this means in practice
- •For robotics engineers: Create control systems for quadruped robots that navigate realistically and safely around moving obstacles in real time.
- •For industrial automation teams: Deploy agile four-legged robots safely in warehouses or factories where obstacles and layouts change dynamically.
Authors
Gage MacLin, Mason Peterson, Camilla Tabasso, Venanzio Cichella, Joshua Migdal, Tye Brady, Taskin Padir
Abstract
Agile quadruped navigation in dynamic environments demands trajectories that are both dynamically feasible and provably collision-free. Most real-time planners achieve tractability through a convex reduction that requires simplified kinematic models, discarding motion that the robot can readily execute, and certify safety only at discretization points, leaving open possible inter-sample collisions with moving obstacles. We present a real-time motion planning framework that retains the fully-actuated $SE(2)$ model of a quadruped, with its anisotropic, heading-coupled velocity bounds, and accepts the resulting nonconvexity. By parameterizing trajectories as composite Bernstein polynomials, we exploit the convex hull property to certify collision avoidance and the input bounds over the entirety of each planning horizon. Solved at online replanning rates, the optimization runs in closed loop, commanding body velocities to the locomotion controller. We validate the approach in simulation against ablations and external baselines, and in hardware experiments on a quadruped navigating among unknown moving obstacles, recovering both lateral and heading-coupled motion that simplified models exclude.