Overcoming Explicit Environment Representations with Geometric Fabrics
Max Spahn, Saray Bakker, Javier Alonso-Mora
AI summary
Problem
Geometric Fabrics traditionally require explicit geometric obstacle models, which are difficult and slow to compute in dynamic, unstructured environments. This creates a bottleneck for real-time reactive robot control.
Approach
The authors map implicit environment data—Signed Distance Fields, Free Space Decomposition, and raw point clouds—into the fabric framework by computing numerical gradients for the necessary pullback operations.
Key results
- Integration of SDFs, FSD, and raw sensor data into Geometric Fabrics
- Derivation of numerical gradient computation for fabric pullback operations
- Real-time reactive control demonstrated on ground robots and manipulators with dynamic obstacles
- Validation through simulations and real-world experiments confirming practical applicability
Why it matters
Provides a computationally efficient pathway for reactive robot navigation and manipulation in complex, changing environments without relying on fragile explicit perception pipelines.
Abstract
Deployment of robots in dynamic environments re- quires reactive trajectory generation. While optimization-based methods, such as Model Predictive Control focus on constraint verificaction, Geometric Fabrics offer a computationally efficient way to generate trajectories that include all avoidance behaviors if the environment can be represented as a set of object primitives. Obtaining such a representation from sensor data is challenging, especially in dynamic environments. In this paper, we integrate implicit environment representations, such as Signed Distance Fields and Free Space Decomposition into the framework of Geometric Fabrics. In the process, we derive how numerical gradients can be integrated into the push and pull operations in Geometric Fabrics. Our experiments reveal that both, ground robots and robotic manipulators, can be controlled using these implicit representations. Moreover, we show that, unlike the explicit representation, implicit representations can be used in the presence of dynamic obstacles without further considerations. Finally, we demonstrate our methods in the real-world, showing the applicability of our approach in practice.