Simulation of aperiodic bipedal sprinting.

Area of Science:

• Biomechanics
• Robotics
• Computational Science

Background:

• Previous dynamic simulations of legged locomotion often used periodicity constraints, suitable for steady-state gaits.
• Sprinting from rest is inherently aperiodic, with rapid acceleration being crucial for performance.

Purpose of the Study:

• To develop a novel simulation for aperiodic sprinting using a modified spring-loaded inverted pendulum (SLIP) biped model.
• To determine optimal control strategies that minimize sprint time over a 20-meter distance.

Main Methods:

• Employed optimal control to minimize sprint time, solving the problem using direct multiple shooting and sequential quadratic programming.
• Utilized an initial 'jogging' simulation with proportional-derivative feedback as a starting point for optimization.
• Developed a modified SLIP biped model for simulating aperiodic locomotion.

Main Results:

• The optimized simulation significantly reduced locomotion time (2.79s vs. 6.64s) compared to the initial guess.
• The model generated substantially greater initial forward impulses (four times higher) for rapid acceleration.
• The simulation replicated key human sprinting behaviors: forward trunk lean, trunk straightening during acceleration, and a finishing dive.

Conclusions:

• This study presents the first optimal control simulation of multistep aperiodic sprinting.
• The optimizer autonomously discovered complex gait features, including foot contact modulation and a finishing dive, without explicit constraints.
• Future work will incorporate musculotendon actuators and joints to investigate musculoskeletal mechanics' influence on gait speed.