Voluntary Retreat for Interference Reduction

Collaboration is a great resource. As proven by biologists, physicists, and a number of social studies, groups of collaborating individuals can address complex tasks even under limited information and skills. Ants cooperate to create a sophisticated network of underground tunnels. Marine bacteria living at great depths build long chains in order to reach lower depths, more abundant with nutrients. Such complex processes are usually accomplished while simultaneously solving other, more elementary, ones. For instance, hidden tasks such as avoid collisions, navigation in the environment, or return home are all examples of behaviors that do not represent a primary goal, yet their presence is of fundamental importance.

Importantly, as the number of collaborating individual increases, the detrimental impact of these secondary behaviors on the performance of the main task increase as well. For example, when increasing the number of people attempting to exit a building through a single door, the number of individuals actually passing through the door per unit of time reduces. Similarly, in densely-packed robot swarms operating in confined regions, spatial interference – which manifests itself as a competition for physical space – forces robots to spend more time navigating around each other rather than performing the primary task.

This aim of this work is to develop a decentralized algorithm that enables individual robots to decide whether to continue participating in their current task and contribute to the overall mission, or vacate the region so as to reduce the negative effects that interference has on the overall efficiency of the swarm.

As shown in the video, this decentralized voluntary retreat process was applied to a distributed collection task. Here, a team of robots collect and deposit objects from one set of locations (pick-up points) to another (drop-off) in the elliptical region. Robots do not communicate and collect only binary information regarding the presence of other robots around them to make the decision to stay or retreat. Given the optimal number of required robots, 5 in this case, individuals decide to retreat when they estimate a higher-than-optimal participation in the task. The algorithm is capable of distributedly adjusting and maintaining the desired number of robots, even when new robots are added at a steady rate.

The solution to this problem provides useful insights on how limited resources can be optimally exploited by a collaborating community of individuals. Practical tasks that can potentially benefit from similar approaches span from distribution of goods, environmental cleaning tasks, and transportation.

Investigators:

  • Siddharth Mayya
  • Pietro Pierpaoli
  • Magnus Egerstedt

Related Publications:

Siddharth Mayya, Pietro Pierpaoli, Magnus Egerstedt, “Voluntary Retreat for Decentralized Interference Reduction in Robot Swarms”, Under Review, IEEE International Conference on Robotics and Automation (ICRA) 2018.

Comments are closed.