Can't Touch This: Real-Time, Safe Motion Planning and Control for Manipulators Under Uncertainty

Jonathan Michaux,Patrick Holmes,Bohao Zhang,Che Chen,Baiyue Wang,Shrey Sahgal,Tiancheng Zhang,Sidhartha Dey,Shreyas Kousik,Ram Vasudevan

2023-11-02

Abstract:Ensuring safe, real-time motion planning in arbitrary environments requires a robotic manipulator to avoid collisions, obey joint limits, and account for uncertainties in the mass and inertia of objects and the robot itself. This paper proposes Autonomous Robust Manipulation via Optimization with Uncertainty-aware Reachability (ARMOUR), a provably-safe, receding-horizon trajectory planner and tracking controller framework for robotic manipulators to address these challenges. ARMOUR first constructs a robust controller that tracks desired trajectories with bounded error despite uncertain dynamics. ARMOUR then uses a novel recursive Newton-Euler method to compute all inputs required to track any trajectory within a continuum of desired trajectories. Finally, ARMOUR over-approximates the swept volume of the manipulator; this enables one to formulate an optimization problem that can be solved in real-time to synthesize provably-safe motions. This paper compares ARMOUR to state of the art methods on a set of challenging manipulation examples in simulation and demonstrates its ability to ensure safety on real hardware in the presence of model uncertainty without sacrificing performance. Project page: <a class="link-external link-https" href="https://roahmlab.github.io/armour/" rel="external noopener nofollow">this https URL</a>.

Robotics,Systems and Control,Optimization and Control

What problem does this paper attempt to address?

The paper attempts to address the problem of how to ensure that a robotic manipulator can safely avoid collisions, adhere to joint limits, and consider the uncertainties in the mass and inertia of objects and the robot itself during real-time motion planning and control in uncertain dynamic environments.

Specifically, the paper proposes a method named ARMOUR (Autonomous Robust Manipulation via Optimization with Uncertainty-aware Reachability) to solve the following issues:

1. **Avoiding Collisions**: Ensuring that the robotic manipulator does not collide with obstacles in the environment during its motion.
2. **Adhering to Joint Limits**: Ensuring that the joint positions, velocities, and torques of the manipulator remain within allowable ranges to avoid damaging the robot.
3. **Handling Uncertainty**: Considering the uncertainties in the mass and inertia parameters of the robot and the objects being manipulated, ensuring safe motion under these uncertain conditions.

ARMOUR achieves these goals through the following key steps:

1. **Constructing a Robust Controller**: Designing a robust controller capable of tracking the desired trajectory, ensuring that the tracking error remains within acceptable bounds even in the presence of dynamic model uncertainties.
2. **Computing Forward Reachable Sets**: Using a novel Recursive Newton-Euler Algorithm (RNEA) to compute all possible inputs to track any desired trajectory.
3. **Optimizing Trajectories**: Transforming the motion planning problem into an optimization problem by over-approximating the swept volume of the manipulator and solving it in real-time to generate safe motion trajectories.

The paper validates the effectiveness of ARMOUR through simulations and real hardware experiments, demonstrating its safety advantages in handling uncertain dynamic conditions.