Stochastic Hedgehog

Uncertainty in Microgravity Hopping

R. Davis Born

Aug to Dec 2016


Overview

Stanford’s Autonomous Systems Lab and NASA’s Jet Propulsion Lab are developing a fully enclosed rover called Hedgehog that hops with internal flywheels. A prototype has been built, and the project is potentially transitioning to the design of a mission vehicle. A primary obstacle is the chaotic hopping motion, which makes Hedgehog difficult to control. Thus, one of the first tasks is to characterize the uncertainty of this hopping motion and attempt to reduce it.

Personal contributions

Hedgehog would ideally stop its motion after a single bounce. Thus, I first investigated the dynamics of different strategies for energy dissipation on impact in one dimension using MATLAB simulations. Simplicity and high energy dissipation were desired.

This 1D motion was largely deterministic, so I then wrote MATLAB simulations in two dimensions to build probability distributions of post-hop state over a range of initial angle of attack and angular rate imposed by the flywheels. Low variance was desired.

Significant effort was put into developing an appropriate model of the impact dynamics implemented in a way that would allow the simulation to converge. A penetration model with a non-penetrating proxy was used with MATLAB’s ODE45 with triggers to effectively model impact impulse and friction effects

Finally, I summarized the most desirable strategies from the one- and two-dimensional simulations to offer recommendations for the project’s direction.

Results / deliverables

  • Conference-style report & presentation detailing methods & findings

  • Identification & summary of promising project directions

  • Simulation package for Hedgehog rover written in MATLAB

  • Statistical analysis package for simulation results written in MATLAB

Software

MATLAB

Latex

MS Office

SKILLS

Dynamics

Statistics

Data VIS

Presentation

Documentation