A Multi-Tool Allocation Approach for Optimized Weed Removal in Autonomous Agriculture

Abstract

Hand weeding has traditionally been a labor-intensive and time consuming task, making it an ideal candidate for automation. However, fields with high weed density remain a challenge for autonomous sys tems. In such scenarios, the limited number of robots and their tooling capacities create bottlenecks, leading to extended mission times. Nuga is a more capable system proposed by Paltech to improve throughput efficiency and reduce total mission duration. These benefits, how ever, depend on effectively solving the task allocation problem with a focus on minimizing tool idle time. To address this, we propose three task allocation algorithms of different paradigms: graph search, optimization-based, and market-based. Our results demonstrate that these approaches can reduce total mission time by up to 22.8% in high density scenarios and decrease tool idle time by as much as 94.9% in comparison with the baseline method.

9