Autonomous Robots: Model Predictive Control

Examine the limits of classic logic in driving tasks with a playground simulator. See how model predictive control predicts trajectories, outputs pedal and steering, and replans to reach the goal.

Learn how to define models and cost functions within model predictive control, explore prediction and optimization, and apply these ideas in assignment zero using a shower temperature example.

Learn how a model predictive control framework uses a plant model and a cost function to optimize knob angles over a prediction horizon, with the optimizer seeking the lowest cost.

Apply model predictive control to drive a car from start to end, using a motion model, cost function, and a prediction horizon to optimize pedal inputs.

Explore assignment one intro in autonomous robots: learn how to set simulator options, understand full recalculate trade-offs, and implement the plant model and cost function for model predictive control.

Walkthrough of an assignment on autonomous robots using model predictive control, showing a plant model with pedal-to-acceleration mapping, state updates, and speed-limit cost tradeoffs across horizons.

Learn how horizon length in model predictive control affects solution quality and computation time, then target two reference positions with practical orientation changes and cost considerations for timely robot execution.

Apply a model predictive control approach to obstacle avoidance by crafting a cost function that penalizes proximity using the inverse of the distance to the obstacle.

Tune horizon and modify the cost function to penalize closeness to the obstacle as you navigate toward the reference goal from the starting point, trying values like 5 or 20.

Deliver the final assignment solution for model predictive control, detailing obstacle avoidance with a 15-step horizon, a smooth distance-based obstacle cost in the cost function.