A Testbed for Multi-Drone Volleyball Game Combining Motion Control and Strategic Play

View PDF file from the paper entitled Vollybots: a TestBed for Multi-Multibly Ballbly Game that combines movement control and strategic play, by Zelai Xu and 11 other books
PDF HTML (experimental) view
a summary:ROBOT SPORTS, which is well -defined goals, explicit rules, and dynamic reactions, offers perfect scenarios to show the embodiment. In this paper, we offer Volleybots, a new robot sport is tested where multiple drones cooperate and compete in volleyball under physical dynamics. Volleybots combines three features within a unified platform: competitive play and cooperation, the rotation -based interaction structure, and the Agile 3D. It challenges competitive and cooperative playing every drone to coordinate with its teammates, with the anticipation and confrontation of the opposition teams. The rotating -based interaction requires accurately, precisely forecasting the state, and managing long -horizontal time. Agile 3D maneuver requires a rapid acceleration, sharp turns, and 3D microorganisms despite the low quadruple dynamics. These interlocking features result in a complex problem that combines movement control and strategic play, with no expert demonstrations available. We offer a comprehensive set of tasks ranging from mono -bronching exercises to multi -program cooperative and competitive tasks, accompanied by basic assessments of multi -agent representative learning (MARL) and the theoretical algorithms of the game. The results of the simulation show that the methods of learning to strengthen quality (RL) outperform the methods of quality in the tasks of one agent, but both the rituals fight in the complex tasks that combine movement control and strategic play. In addition, we design a hierarchical policy that achieves a 69.5 % victory rate compared to the strongest baseline in mission 3 against 3, which confirms its potential as an effective solution to address the complex interaction between low -level and high -level strategy. The project page at the URL https.
The application date
From: Ruiz Chang [view email]
[v1]
Tuesday, 4 February 2025 02:07:23 UTC (14,076 KB)
[v2]
Tuesday, 11 February 2025 03:00:12 UTC (14,076 KB)
[v3]
Saturday, 17 May 2025 11:20:39 UTC (9,225 KB)
[v4]
Tuesday, 8 July 2025 13:30:59 UTC (9,225 KB)
Don’t miss more hot News like this! Click here to discover the latest in AI news!
2025-07-09 04:00:00