Coach planning with opponent models for distributed execution

Riley, Patrick; Veloso, Manuela

doi:10.1007/s10458-006-7449-z

Cited by 7 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In terms of guiding agents with high-level commands, our work is similar to the coaching system in RoboCup soccer simulation where a coach who gets an overview of the whole game sends commands to the players of its own team. However, the coach is also a software agent (there is no human in the loop) and its decision is mainly on recognizing and selecting the opponent model [12,17]. For planning with human guidance, MAPGEN [1], the planning system for Mars rover missions, allows operators to define constraints and rules for a plan, which are subsequently enforced by automated planners to produce the plan.…”

Section: Related Workmentioning

confidence: 99%

Multi-agent Planning with High-Level Human Guidance

Zilberstein

Jennings

2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Planning and coordination of multiple agents in the presence of uncertainty and noisy sensors is extremely hard. A human operator who observes a multi-agent team can provide valuable guidance to the team based on her superior ability to interpret observations and assess the overall situation. We propose an extension of decentralized POMDPs that allows such human guidance to be factored into the planning and execution processes. Human guidance in our framework consists of intuitive high-level commands that the agents must translate into a suitable joint plan that is sensitive to what they know from local observations. The result is a framework that allows multi-agent systems to benefit from the complex strategic thinking of a human supervising them. We evaluate this approach on several common benchmark problems and show that it can lead to dramatic improvement in performance.

show abstract

Section: Related Workmentioning

confidence: 99%

Multi-agent Planning with High-Level Human Guidance

Zilberstein

Jennings

2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Visual information is sent six or seven times per second. During a standard 10-minute game, this gives 6,000 action chances and 4,000 receipts of visual information [12]. Based on the scene-information, each agent selects an action with the following steps: 1) evaluating the subjective state described in world model; 2) making a decision using the strategy algorithm; and 3) generating and refining basic commands (such as dashing in a given direction with certain power, turning the body or neck at an angle, kicking the ball at an angle with specified power, or slide-tackling the ball).…”

Section: Introductionmentioning

confidence: 99%

“…Visual information is sent six or seven times per second. During a standard 10-minute game, this gives 6,000 action chances and 4,000 receipts of visual information [12].…”

Section: Introductionmentioning

confidence: 99%

Offensive Strategy in the 2D Soccer Simulation League Using Multi-Group Ant Colony Optimization

Chen

2016

International Journal of Advanced Robotic Systems

View full text Add to dashboard Cite

The 2D soccer simulation league is one of the best test beds for the research of artificial intelligence (AI). It has achieved great successes in the domain of multi-agent cooperation and machine learning. However, the problem of integral offensive strategy has not been solved because of the dynamic and unpredictable nature of the environment. In this paper, we present a novel offensive strategy based on multi-group ant colony optimization (MACO-OS). The strategy uses the pheromone evaporation mechanism to count the preference value of each attack action in different environments, and saves the values of success rate and preference in an attack information tree in the background. The decision module of the attacker then selects the best attack action according to the preference value. The MACO-OS approach has been successfully implemented in our 2D soccer simulation team in RoboCup competitions. The experimental results have indicated that the agents developed with this strategy, along with related techniques, delivered outstanding performances.

show abstract

“…General replanning algorithms are popular within case-based learning [8], [9], although they do not scale well in control problems with multiple actors and increasing numbers of environmental features. Replanning within spatial domains is frequently considered in the RoboCup domain [10], [11]. Plan adaptation can be seen as an 'off-line' version of cooperative control [12], where research efforts are concentrating on maintaining a specific aspect of team behavior (like the team formation, see e.g.…”

Section: Introductionmentioning

confidence: 99%

Multi-robot plan adaptation by constrained minimal distortion feature mapping

Takács¹,

Demiris²

2009

2009 IEEE International Conference on Robotics and Automation

View full text Add to dashboard Cite

We propose a novel method for multi-robot plan adaptation which can be used for adapting existing spatial plans of robotic teams to new environments or imitating collaborative spatial teamwork of robots in novel situations. The algorithm selects correspondences between previous and current spatial features by the application of pairwise constraints, and generates the transformation function with a fast regular grid approximation which minimizes distortion. The algorithm requires minimal domain knowledge, is capable of transforming the spatial aspects of collaborative team behavior and performs better in noisy problems with large displacements than the most generally used quadratic differences method. The algorithm can be utilized for rapid plan adaptation, plan generalization or team behavior imitation. Methods are demonstrated on a multi-robot control problem in a random environment.

show abstract

Coach planning with opponent models for distributed execution

Cited by 7 publications

References 37 publications

Multi-agent Planning with High-Level Human Guidance

Multi-agent Planning with High-Level Human Guidance

Offensive Strategy in the 2D Soccer Simulation League Using Multi-Group Ant Colony Optimization

Multi-robot plan adaptation by constrained minimal distortion feature mapping

Contact Info

Product

Resources

About