Generation of multiagent animation for object transportation using deep reinforcement learning and blend‐trees

Abstract

This paper proposes a framework that integrates reinforcement learning and blend-trees to generate animation of multiple agents for object transportation. The main idea is that in the learning stage, policies are learned to control agents to perform specific skills, including navigation, pushing, and orientation adjustment. The policies determine the blending parameters of the blend-trees to achieve locomotion control of the agents. In the simulation stage, the policies are combined to control the agents to navigate, push objects, and adjust orientation of the objects. We demonstrated several examples to show that the framework is capable of generating animation of multiple agents in different scenarios.

Animation generation for object transportation with a rope using deep reinforcement learning

“…To avoid colliding with dynamic obstacles, agents rely on collision fields to perform collision prediction and plan beforehand their movements in cluttered environments 8 . Blend‐trees can be adopted to enhance the character animations while the characters push objects to goal positions 14 . In cart‐pulling animations, two agents use a rope to pull a cart.…”

Generation of cart‐pulling animation in a multiagent environment using deep learning

“…22 Deep learning is integrated with blend-trees for generating natural locomotion animation of characters in object transportation. 12…”

“…To achieve reciprocal collision avoidance, agents learn to move while avoid collision with each other 22 . Deep learning is integrated with blend‐trees for generating natural locomotion animation of characters in object transportation 12 …”

“…For example, policies can be learnt to control agents to perform pulling and pushing objects in animation generation 11 . Deep learning techniques can be integrated with blend‐trees for generating object transportation animation 12 . Furthermore, multiagents reinforcement learning (MARL) makes agents perform collaboration or interaction behaviors to enrich the content of animation 13 .…”

“…11 Deep learning techniques can be integrated with blend-trees for generating object transportation animation. 12 Furthermore, multiagents reinforcement learning (MARL) makes agents perform collaboration or interaction behaviors to enrich the content of animation. 13 In this article, we propose a reinforcement-based framework that enables two workers to pull a cart with two ropes.…”