Optimization-Based Motion Prediction in Biomechanical Systems

Optimization-based motion prediction in biomechanical systems integrates mathematical models of the human musculoskeletal apparatus with algorithmic techniques to forecast movement trajectories under given tasks and constraints. By formulating motion as the solution of an optimisation problem, researchers aim to identify joint angles, muscle activations and external forces that minimise or balance criteria such as metabolic cost, joint discomfort or mechanical effort. This paradigm draws upon inverse dynamics to compute internal loads from desired motions and inverse kinematics to derive the joint configurations necessary for end-effector placement. Advances in computational power and solver efficiency have enabled more elaborate representations of anatomical structures and incorporation of subject-specific parameters. The approach finds applications in clinical gait analysis, ergonomic assessment, rehabilitation device control and the design of human-centred robots. Recent efforts also explore hybrid frameworks that combine optimisation with data-driven models to enhance real-time prediction and to accommodate variability across individuals. The global significance of this field lies in its potential to improve patient outcomes, reduce workplace injury risk and advance the naturalness of robot-mediated interactions.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

A subtask-based inverse dynamic optimisation formulation has been developed to predict two-dimensional box delivery motions. By dividing the task into lifting, transitions, carrying and unloading subtasks, each governed by tailored cost functions—such as joint torque squared and a combination of discomfort and torque—researchers achieved close agreement with experimental joint-angle profiles. This method proved computationally efficient and offers practical guidance for reducing injury risk during manual handling.

An analytic inverse kinematic programme for human upper-limb posture prediction during reaching tasks employed both joint displacement and discomfort functions. Upon finding that single-criterion objectives fell short of replicating natural postures, a bi-criterion cost function was introduced and tuned via golden section search. The resulting model delivered substantially improved accuracy in reproducing observed arm postures.

An inverse-optimisation approach has been applied to map human dual-arm manipulation skills onto a humanoid robot. By assuming that human motion adheres to an optimality principle, multiple criterion functions—kinetic energy, joint velocity, ergonomic proximity and manipulability—were weighted through an inverse algorithm. The derived weight set enabled the robot to perform tasks such as drawer opening with human-like fluidity and coordination.

Optimization-Based Motion Prediction in Biomechanical Systems publication trend

The graph below shows the total number of articles in optimization-based motion prediction in biomechanical systems across all publications each year (not limited to Nature Index journals).

Technical terms

Inverse dynamics: Calculation of internal forces and torques required to produce a given motion.

Inverse kinematics: Determination of joint configurations that achieve a target end-effector position.

Cost function: A quantitative metric to be minimised or maximised in an optimisation problem.

Musculoskeletal model: A computational representation of bones, joints and muscles for simulating motion.

Manipulability: A measure of how readily a kinematic chain can move or exert forces in various directions.

Subtask-based optimisation: Division of a complex motion into segments, each optimised under specific criteria.

Bi-criterion objective: An optimisation goal combining two distinct performance measures for improved fidelity.

References

1. Two-Dimensional Symmetric Box Delivery Motion Prediction and Validation: Subtask-Based Optimization Method. Applied Sciences (2020).
2. Determining human upper limb postures with a developed inverse kinematic method. Robotica (2022).
3. Toward optimal mapping of human dual-arm motion to humanoid motion for tasks involving contact with the environment. International Journal of Advanced Robotic Systems (2018).