← ArchivePaper2014
Generalizing Locomotion Style to New Animals with Inverse Optimal Regression
Inverse optimal control framework for generalizing locomotion style to novel animal morphologies by regressing locomotion objectives.
Abstract
We present a technique for analyzing a set of animal gaits to predict the gait of a new animal from its shape alone. This method works on a wide range of bipeds and quadrupeds, and adapts the motion style to the size and shape of the animal. We achieve this by combining inverse optimization with sparse data interpolation. Starting with a set of reference walking gaits extracted from sagittal plane video footage, we first use inverse optimization to learn physically motivated parameters describing the style of each of these gaits. Given a new animal, we estimate the parameters describing its gait with sparse data interpolation, then solve a forward optimization problem to synthesize the final gait. To improve the realism of the results, we introduce a novel algorithm called joint inverse optimization which learns coherent patterns in motion style from a database of example animal-gait pairs. We quantify the predictive performance of our model by comparing its synthesized gaits to ground truth walking motions for a range of different animals. We also apply our method to the prediction of gaits for dinosaurs and other extinct creatures.
How to read this
- Category
- Method: inverse-optimal-control synthesis of animal locomotion style
- Contributions
- A framework that predicts a new animal's gait from its shape alone, adapting motion style to size and shape across bipeds and quadrupeds
- Inverse optimization to learn physically motivated style parameters from reference gaits, combined with sparse data interpolation to estimate parameters for a new animal
- A joint inverse optimization algorithm that learns coherent style patterns from a database of animal-gait pairs, with application to extinct creatures
- Context
- Rooted in physics-based optimization of motion (Witkin and Kass's 'Spacetime Constraints'), recast as inverse optimal control to recover the objectives that explain observed gaits and regress them to new morphologies.Builds on: Spacetime Constraints
- Correctness
- Predictive performance is quantified against ground-truth walking motions for a range of animals; extrapolation to dinosaurs and extinct creatures is plausible by construction but inherently unverifiable, and inputs are reference gaits from sagittal-plane video, so out-of-plane motion is not directly modeled.
- Clarity
- Conceptually clear but technically dense; a first pass conveys the inverse-then-forward optimization loop, and a second pass is worth it for the joint inverse optimization formulation.
- How to read it
- Read for the inverse-optimization-plus-interpolation pipeline; focus on what the learned style parameters represent and how the joint formulation pools across examples, with a careful second pass before trusting the extinct-animal extrapolations.
Builds on
Built upon by
Nothing yet.
Related work
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →