Simulated Annealing for Multi-Robot Ergodic Information Acquisition Using Graph-Based Discretization
Benjamin Wong, Aaron Weber, Mohamed Safwat, Santosh Devasia, Ashis Banerjee
AI summary
Problem
Traditional ergodic control requires a known target sampling distribution, but in active information acquisition, region noise levels are initially unknown and unreliable, causing fluctuating distributions and inefficient resource allocation.
Approach
The method applies simulated annealing to gradually shift the ergodic target distribution from uniform to variance-proportional by adjusting a Boltzmann coldness parameter, solved via graph-based REMC planning.
Key results
- Substantially improved transient and asymptotic sampling entropy
- Smooth transition from uniform exploration to optimal allocation
- Validated on a physical TurtleBot swarm system
- Robust performance despite unreliable initial variance estimates
Why it matters
Provides a practical, adaptive planning framework for multi-robot teams conducting surveillance, inspection, or disaster response in uncertain environments.
Abstract
One of the goals of active information acquisition using multi-robot teams is to keep the relative uncertainty in each region at the same level to maintain identical acquisition quality (e.g., consistent target detection) in all the regions. To achieve this goal, ergodic coverage can be used to assign the number of samples according to the quality of observation, i.e., sampling noise levels. However, the noise levels are unknown to the robots. Although this noise can be estimated from samples, the estimates are unreliable at first and can generate fluctuating values. The main contribution of this paper is to use simu- lated annealing to generate the target sampling distribution, starting from uniform and gradually shifting to an estimated optimal distribution, by varying the coldness parameter of a Boltzmann distribution with the estimated sampling entropy as energy. Simulation results show a substantial improvement of both transient and asymptotic entropy compared to both uniform and direct-ergodic searches. Finally, a demonstration is performed with a TurtleBot swarm system to validate the physical applicability of the algorithm.