Autonomy Playground
Interactive apps on guidance, navigation and control of autonomous vehicles. Each app lets you drag, change a gain, run it again, and watch what happens. Everything runs in your browser.
On YouTube: short narrated lessons with simulations of surface vessels, underwater vehicles and drones.
▶ The YouTube channelGuidance
LOS Guidance
Line-of-sight (LOS) guidance for following a waypoint-based path with smooth turns. Drag the waypoints while two vessels follow the path, and compare two ways of choosing the reference point on the path.
Collision avoidance
Simulation-Based MPC
Model predictive control (MPC) on an Otter uncrewed surface vehicle: a fan of turn-rate plans, predicted with a model and scored by a cost, the plans that come too close to a kayak or a buoy dropped, and the cheapest one carried out for one step. Change the horizon, the cost and the plans, and drag the kayak across the path.
Simulation-Based MPC in 3D
The same idea in 3D, on a quadrotor: a set of acceleration plans, predicted and scored, the ones that come too close to another drone, an obstacle or the ground dropped. It can climb over, go round, or brake and wait. Move the other drone, balloons and pillars in 3D while it flies.
State estimation
State Observer
A ball and beam with a noisy distance sensor. Estimate the ball's velocity by finite differences, a filtered derivative or a Luenberger observer, choose which one the controller uses, and tune the observer's gains.