Advanced Skills through Multiple Adversarial Motion Priors in Reinforcement Learning

Vollenweider, Eric; Bjelonic, Marko; Klemm, Victor; Rudin, Nikita; Lee, Joonho; Hutter, Marco

doi:10.1109/icra48891.2023.10160751

Cited by 15 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The complexity of the reward function often increases with the complexity of the task. For example, some previous studies have designed detailed reward terms to approximate the robot's actions to teacher samples through adversarial and imitation learning [3,6]. For a standing task, the reward function needs to be designed as follows: R = r α + r height + r feet + r wheels + r shoulder + r stand pose (11) where…”

Section: Rewardsmentioning

confidence: 99%

“…By allowing the robot to learn through trial-and-error interactions with its environment, RL can discover optimal control policies without relying on explicit programming. Recent advancements in deep reinforcement learning (DRL) have further enhanced the capabilities of quadrupedal robots, enabling them to learn complex behaviors such as jumping, backflips, and even bipedal walking [4][5][6]. Enabling quadrupedal robots to walk on three or two legs has significant application value, as it allows the free legs of quadrupedal robots to perform manipulation tasks, combining the advantages of both quadrupedal and humanoid robots.…”

Section: Introduction 1backgroundmentioning

confidence: 99%

See 1 more Smart Citation

Learning Advanced Locomotion for Quadrupedal Robots: A Distributed Multi-Agent Reinforcement Learning Framework with Riemannian Motion Policies

Wang,

Sagawa,

Yoshiyasu

2024

Robotics

View full text Add to dashboard Cite

Recent advancements in quadrupedal robotics have explored the motor potential of these machines beyond simple walking, enabling highly dynamic skills such as jumping, backflips, and even bipedal locomotion. While reinforcement learning has demonstrated excellent performance in this domain, it often relies on complex reward function tuning and prolonged training times, and the interpretability is not satisfactory. Riemannian motion policies, a reactive control method, excel in handling highly dynamic systems but are generally limited to fully actuated systems, making their application to underactuated quadrupedal robots challenging. To address these limitations, we propose a novel framework that treats each leg of a quadrupedal robot as an intelligent agent and employs multi-agent reinforcement learning to coordinate the motion of all four legs. This decomposition satisfies the conditions for utilizing Riemannian motion policies and eliminates the need for complex reward functions, simplifying the learning process for high-level motion modalities. Our simulation experiments demonstrate that the proposed method enables quadrupedal robots to learn stable locomotion using three, two, or even a single leg, offering advantages in training speed, success rate, and stability compared to traditional approaches, and better interpretability. This research explores the possibility of developing more efficient and adaptable control policies for quadrupedal robots.

show abstract

Section: Rewardsmentioning

confidence: 99%

Section: Introduction 1backgroundmentioning

confidence: 99%