Biomimetic brain machine interfaces for the control of movement

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Current brain machine interfaces (BMIs) enable real-time control of external devices using neural signals but largely disregard musculoskeletal system dynamics and somatosensory feedback.
• This limitation hinders the restoration of natural limb movement for individuals with motor impairments.

Purpose of the Study:

• To discuss novel brain machine interface (BMI) approaches inspired by sensorimotor physiology for more natural limb movement control.
• To explore advancements in decoder development, real-time dynamical signal prediction, and the integration of somatosensory feedback.

Main Methods:

• Developing advanced decoders using structured, nonlinear models and incorporating limb state information.
• Implementing real-time prediction and control of dynamical signals (joint torque, force, EMG) for physical systems mimicking limb dynamics.
• Investigating somatosensory feedback integration, including its benefits and neural representations.

Main Results:

• Proposed alternative decoder methods for improved BMI performance, especially for paralyzed individuals.
• Demonstrated real-time prediction and control of limb-like dynamics.
• Identified critical factors for effective somatosensory feedback incorporation.

Conclusions:

• Novel BMI strategies incorporating musculoskeletal dynamics and somatosensory feedback are crucial for restoring natural limb movement.
• Advancements in decoding, dynamical control, and sensory feedback promise more intuitive and effective BMIs for motor rehabilitation.