Sometimes, one robot is not enough.
Consider a search and rescue mission to find a traveler lost in the jungle. Rescuers may want to deploy a squad of wheeled robots to roam the forest, perhaps with the help of a drone scouring the scene from above. The benefits of the robot team are obvious. But revolving that team is not an easy thing. To ensure that robots are not replicating each other’s efforts or wasting energy on a complex search trajectory?
MIT researchers have designed an algorithm to ensure the fruitful collaboration of robot teams collecting information. Their approach relies on balancing a tradeoff between the data collected and the energy – which eliminates the possibility that a robot may perform a useless maneuver to gain only one type of information. . Researchers say this assurance is critical to the success of robot teams in complex, unpredictable environments. “Our method provides comfort, because we know it will not fail, thanks to the worst-case performance of the algorithm,” a Ph.D. Student in MIT’s Department of Aeronautics and Astronauts (AeroAstro).
Will be presented in research IEEE International Conference on Robots and Automation In May. Cai is the lead author of the paper. His co-authors include Jonathan McHorlin, RC McLaurin Professor of Aeronautics and Astronautics at MIT; Brent Schlotfelt and George J. Pappas, both of the University of Pennsylvania; And Nikolay Atanasov of the University of California at San Diego.
Robot teams often rely on an overreaching rule to gather information: the higher the merger. “The assumption is that it never hurts to collect more information,” Cai says. “If there is a fixed battery life, use it to get as much mileage as possible.” This objective is often executed sequentially – each robot evaluates the situation and plans its trajectory one after the other. This is a straightforward process, and it generally works well when information is the sole purpose. But problems arise when energy efficiency becomes a factor.
Cai says that the advantages of gathering additional information often diminish over time. For example, if you already have 99 photos of a jungle, it may not be worth sending a robot on a mile-long quest to find the 100th picture. “We want to be conscious of the trade-off between information and energy,” Cai says. “It’s not always good to rotate more robots. It can actually be worse when you’re a factor of energy cost.”
The researchers developed a robotic team planning algorithm that optimizes the balance between energy and information. The “objective function” of the algorithm, which determines the value of the robot’s proposed function, is responsible for the gathering of additional information and the diminishing benefits of rising energy costs. Unlike prior planning methods, it does not provide tasks to the robots sequentially. “It’s more of a collaborative effort,” Cai says. “Robots themselves come up with team plans.”
Cai’s method, called Distributed Local Search, is an iterative approach that improves team performance by adding or removing individual robot trajectories from the overall plan of the group. First, each robot independently generates a set of possible trajectories that it can pursue. Subsequently, each robot proposes its trajectory to the rest of the team. The algorithm then accepts or rejects each individual proposal, depending on whether it increases or decreases the objective function of the team. “We allow the robot to plan its own trajectory,” Cai says. “Only when they need to come up with the team’s plan, we let them negotiate. Therefore, it is a distributed delivery.”
Distributed local search proved its precision in computer simulation. The researchers ran their algorithm against competitors in coordination with a simulated team of 10 robots. While distributed local search took slightly more computation time, it guaranteed to successfully complete the robot’s mission, ensuring that no team member was found in a useless expedition for minimal information. “It’s a more expensive method,” Cai says. “But we gain performance.”
According to Geoff Hallinger, a robotic expert at Oregon State University, one day the robot team could help solve real energy information-gathering problems, a robotics expert at Oregon State University. “These techniques apply where the robot team is required to trade between sensing quality and energy expenditure. This will include aerial surveillance and ocean surveillance.”
Cai also points to potential applications in mapping and search-and-rescue activities that rely on efficient data collection. “Improving this inherent ability to gather information will be quite effective,” he says. The researchers next plan to test their algorithms on robot teams in the lab, which includes a mixture of drones and wheeled robots.
Path planning techniques for multiple robots in flexible structures
Non-Monotone Energy-Aware Info Gathering for Heterogeneous Robot Teams. arxiv.org/abs/2101.11093
This story is published courtesy of MIT News (web.mit.edu/newsoffice/), A popular site that covers news about MIT research, innovation and teaching.
Quotes: To help robots work (2021, 14 May) Retrieved 14 May 2021 from https://techxplore.com/news/2021-05-robots-collaborate-job.html
This document is subject to copyright. No part may be reproduced without written permission, except for any impartial behavior for the purpose of private study or research. The content is provided for information purposes only.