Occlusion-Based Cooperative Transport for Concave Objects with a Swarm of Miniature Mobile Robots

TL;DR

This work tackles decentralized collective transport of concave objects by first converting them into a convex pseudo object through concavity filling, and then applying an occlusion-based transport strategy. The two-stage approach requires no prior object geometry or inter-agent communication, making it robust to initial configurations. Experiments in Buzz and ARGoS validate both concavity filling and subsequent convex-object transport, demonstrating viability across multiple shapes and random initial placements. Limitations include proximity sensing range and lack of onboard vision, with future work aiming to incorporate cameras and integrated end-to-end optimization.

Abstract

An occlusion based strategy for collective transport of a concave object using a swarm of mobile robots has been proposed in this paper. We aim to overcome the challenges of transporting concave objects using decentralized approach. The interesting aspect of this task is that the agents have no prior knowledge about the geometry of the object and do not explicitly communicate with each other. The concept is to eliminate the concavity of the object by filling a number of robots in its cavity and then carry out an occlusion based transport strategy on the newly formed convex object or "pseudo object". We divide our work into two parts: concavity filling of various concave objects and occlusion based collective transport of convex objects.

Figures (10)

• Figure 1: Occlusion-Based Collective Transport for Concave Object
• Figure 2: When robot approaches from convex side
• Figure 3: When robot approaches from concave side
• Figure 4: Experimental Objects
• Figure 5: Experimental Setup: Proposed Filling to form Convex Object