Presentation
Reinforcement Learning Based Decision Support for Adaptive Pacing Control of Hand Activity Level
SessionWednesday Poster Session
DescriptionManual material handling tasks frequently cause hand and wrist musculoskeletal disorders with repetitive motions and improper pacing. This study introduces a reinforcement learning (RL) framework to optimize adaptive pacing control by modeling Hand Activity Level (HAL) as a Markov Decision Process. We evaluated three model-free algorithms: Q-learning, SARSA, and DQN to dynamically adjust work pace and schedule micro-breaks, balancing productivity against ergonomic risk. Results reveal distinct trade-offs; DQN maximized productivity but incurred higher cumulative exposure, while Q-learning minimized risk at the cost of pacing stability. These findings demonstrate RL’s potential to jointly optimize worker safety and operational efficiency.
Event Type
Poster
TimeWednesday, October 21st5:30pm - 6:30pm PDT
Location
Similar Presentations