DiffCo: Auto-Differentiable Proxy Collision Detection with Multi-class Labels for Safety-Aware Trajectory Optimization

Abstract

The objective of trajectory optimization algorithms is to achieve an optimal collision-free path between a start and goal state. In real-world scenarios where environments can be complex and non-homogeneous, a robot needs to be able to gauge whether a state will be in collision with various objects in order to meet some safety metrics. The collision detector should be computationally efficient and, ideally, analytically differentiable to facilitate stable and rapid gradient descent during optimization. However, methods today lack an elegant approach to detect collision differentiably, relying rather on numerical gradients that can be unstable. We present DiffCo, the first, fully auto-differentiable, non-parametric model for collision detection. Its non-parametric behavior allows one to compute collision boundaries on-the-fly and update them, requiring no pre-training and allowing it to update continuously in dynamic environments. It provides robust gradients for trajectory optimization via backpropagation and is often 10-100x faster to compute than its geometric counterparts. DiffCo also extends trivially to modeling different object collision classes for semantically informed trajectory optimization.

Type-2 Fuzzy Model-Based Movement Primitives for Imitation Learning

“…None of them can tackle collisions in an environment that is unstructured, non-convex, and detected with noisy outliers. To handle complex environments, several studies, like [32], [33], utilize the collision detector (e.g. Flexible Collision Library [34]) to provide mesh representations of the environment, which, however, cannot be generated from local observation.…”

Chance-Constrained Motion Planning using Modeled Distance-to-Collision Functions

“…Many methods are proposed that use geometric sensors and modeling techniques to estimate the distance-tocollision [19], [20], and for a brief review, readers can refer to [21], but none considers the measurement models in unstructured environments. Das and Yip [21] proposed one of the first techniques that included uncertainties to the distance measure model.…”