Fastron: An Online Learning-Based Model and Active Learning Strategy for Proxy Collision Detection

TL;DR

Fastron is an online learning-based model that provides fast, conservative proxy collision detection for robotic configuration spaces, significantly reducing computation time while maintaining accuracy, without relying on GPU acceleration.

Contribution

The paper introduces Fastron, a novel online learning algorithm with active learning for efficient proxy collision detection in robotics, outperforming traditional methods in speed and scalability.

Findings

Fastron achieves at least 2x faster checks than polyhedral collision detectors.

Fastron is at least 8x faster than high-precision collision checkers.

Proxy collision checks scale well with increasing obstacles without GPU use.

Abstract

We introduce the Fastron, a configuration space (C-space) model to be used as a proxy to kinematic-based collision detection. The Fastron allows iterative updates to account for a changing environment through a combination of a novel formulation of the kernel perceptron learning algorithm and an active learning strategy. Our simulations on a 7 degree-of-freedom arm indicate that proxy collision checks may be performed at least 2 times faster than an efficient polyhedral collision checker and at least 8 times faster than an efficient high-precision collision checker. The Fastron model provides conservative collision status predictions by padding C-space obstacles, and proxy collision checking time does not scale poorly as the number of workspace obstacles increases. All results were achieved without GPU acceleration or parallel computing.