Optimizing the Sensor Placement for Foot Plantar Center of Pressure without Prior Knowledge Using Deep Reinforcement...

Area of Science:

• Biomechanics
• Machine Learning
• Sensor Technology

Background:

• Traditional foot plantar sensor placement relies on anatomical knowledge, which can be suboptimal.
• Optimizing sensor placement is crucial for accurate center of pressure (COP) measurement during dynamic activities like running.

Discussion:

• A novel deep reinforcement learning (DRL) approach was developed to determine optimal foot plantar sensor placement.
• The DRL agent was trained in a custom environment to minimize the error between the predicted and ground truth COP trajectory.
• The DRL agent explored over 116 quadrillion possible sensor configurations to find the optimal placement.

Key Insights:

• The DRL method successfully generated sensor placements with low mean square error for COP trajectory fitting.
• The algorithm robustly identified optimal sensor placements, demonstrating its effectiveness in a vast search space.
• The DRL approach is adaptable and can be applied to various tasks beyond running, showcasing its versatility.

Outlook:

• Future research could explore DRL for optimizing sensor placement in other biomechanical applications.
• Investigating the interpretability of the learned sensor placements could provide new anatomical insights.
• Real-world validation of DRL-derived sensor placements in diverse populations and activities is warranted.