Learning active tactile perception through belief-space control
This addresses the challenge for robots operating in open worlds to autonomously sense object properties, though it is incremental as it builds on existing tactile perception and control techniques.
The paper tackles the problem of robots estimating unknown physical properties of novel objects through tactile interaction, proposing a method that learns exploration policies using a generative world model and Bayesian filtering, and demonstrates its effectiveness in simulated tasks and on a real robot for height estimation.
Robots operating in an open world will encounter novel objects with unknown physical properties, such as mass, friction, or size. These robots will need to sense these properties through interaction prior to performing downstream tasks with the objects. We propose a method that autonomously learns tactile exploration policies by developing a generative world model that is leveraged to 1) estimate the object's physical parameters using a differentiable Bayesian filtering algorithm and 2) develop an exploration policy using an information-gathering model predictive controller. We evaluate our method on three simulated tasks where the goal is to estimate a desired object property (mass, height or toppling height) through physical interaction. We find that our method is able to discover policies that efficiently gather information about the desired property in an intuitive manner. Finally, we validate our method on a real robot system for the height estimation task, where our method is able to successfully learn and execute an information-gathering policy from scratch.