Skip to main content
Interactive Robotics Laboratory
Yu Gu, Professor

Swarm Foraging

Swarm foraging - economics based resoruce gathering for planetary exploration

The primary focus of this research is to explore and harness the advantages of autonomous robotic swarms over traditional, centralized robotic systems. Our objective is to address a crucial question: how can we use simple, local behaviors to build complex, reliable, and highly effective swarms that excel in a wide range of applications? These applications span from search and rescue operations to planetary exploration and resource gathering. Intriguingly, our work draws on principles from economics, particularly the mechanics of auctions and the concept of opportunity cost, to optimize task allocation and enhance the swarm's effectiveness.

The implications of successful swarm robotics are extensive and transformative. Traditional robotics rely on a centralized framework where specific agents play pivotal roles in the effectiveness of the entire system. This makes them vulnerable; the loss or malfunction of a single agent could severely impact the entire operation. In contrast, robotic swarms, built on distributed frameworks, are robust and resilient. But how do we make these systems not just reliable, but also efficient? Enter economics. By employing economic theories such as auctions to facilitate task allocation, we introduce a sense of 'market competition' among the agents, optimizing their roles based on real-time conditions. This allows for a more fluid, adaptable system that maximizes its operational utility.

Our empirical studies have produced remarkable findings that echo economic principles. When food resources were clustered together, trading behaviors between individual agents—akin to a miniaturized marketplace—led to a substantial increase in food gathered. Agents appeared to weigh the opportunity cost of trading versus foraging, making more efficient choices that collectively benefited the swarm. However, the landscape changed when food was scattered. Here, the opportunity cost of trading outweighed the benefits, leading to a reduction in efficiency. These fluctuations provided a live demonstration of how economic theories like opportunity cost can be pivotal in understanding and improving swarm behavior. Moreover, our experiments also highlighted emergent heterogeneous task specialization among agents. Some agents specialized in acquisition and trading, echoing roles you might find in an economic system, while others focused on exploration and resource discovery. These findings deepen our understanding of swarm robotics, showing that the incorporation of economic theories can significantly boost the adaptability and resilience of these systems.