Manoj Srinivasan

The paper shows that dice stacking, a dynamic manipulation task, can be performed open-loop without state feedback, using simulations and a physical robot.

This study shows that stride-to-stride variability in human walking increases metabolic costs and that simulations neglecting this variability may misestimate metabolic costs.

The paper applies the James-Stein estimator to kinematic and metabolic human data to improve accuracy and reduce sample requirements.

This paper models locomotor adaptation as interactions between a stabilizing controller and a reinforcement learner, predicting adaptation in various scenarios and capturing learning phenomena.

This study investigates how humans control force by examining trade-offs between error reduction and biases in force tracking tasks.

The paper investigates loss functions humans optimize for regression and classification tasks using inverse optimization, finding different exponents and functions describe human behavior under varying data conditions.