Picked from the software, topic, and level of “Class Autonomous Car MPC Path Tracking #6: Testing on Curved Paths, Speed Variations, and Error Evaluation”.
At most one email a week. Unfollow anytime with one click.
Transcript
The full narration of this video. The current sentence is highlighted while the video plays; click any sentence to jump there.
Show transcript
Introduction
Here is the final result, three controllers are tested on a 300-meter winding circuit at three speeds. MPC is the most accurate at all speeds, up to almost 10 times more accurate than pure pursuit. We also see the cost MPC pays at high speeds. Let's start. Hello, welcome to the final episode of the MPC Path Tracking autonomous vehicle series. Ice turns are still relatively easy. Today we test the three controllers on a winding circuit with turns of only 10.7 meters in radius, at speeds of 5, 10, and 15 meters per second, then evaluate the error and steering smoothness as in a scientific paper.
Step 1: Three controllers x three speeds
Create the final test underscore.py file. For each speed and each controller, we run a simulation and record four metrics. Lateral RMS error, maximum lateral error, angular RMS error, and steering smoothness, which is the average change in steering angle per step. Run. At 5 meters per second, MPC has an RMS error of only 0.006 meters, 9.7 times more accurate than pure pursuit. At 10 meters per second, MPC 0.038 meters, Stanley 0.111, and pure pursuit 0.136. At 15 meters per second, MPC 0.079 meters, Stanley 0.107, and pure pursuit deviates up to 1.1 meters at the sharpest turn.
Step 2: Path, error & bar chart
Continue with the graph, third path of the controller on the circuit at 15 meters per second, lateral error over time, and bar chart RMS per speed. Let's look at the graph. On the circuit scale, all three appear to follow the path, but the error graph shows pure pursuit jumping at the sharp turn around the 12th second. The bar chart shows all controllers' errors increasing with speed, and MPC is always the lowest. But at 15 meters per second, MPC steering changes more frequently, 0.55 degrees per step. A 10.7-meter turn at that speed requires lateral acceleration more than twice gravity, beyond the capability of a real car, so MPC works hard at the steering rate limit.
Done!
Done, and the series is complete. We have built a complete research on autonomous vehicle path tracking, bicycle model, reference path, pure pursuit and Stanley, linearization and formulation of MPC, implementation with CVXPi, up to circuit testing at various speeds. For further research, try the tire force dynamics model, nonlinear MPC with Casadi, automatic speed setting based on curvature, or obstacle avoidance. Thank you for following this series. Currently working on a thesis, dissertation, or research in the field of autonomous vehicles and control systems? Cerdas Inovasi Digital helps implement control systems for vehicles, from bicycle model, reference path, pure pursuit and Stanley, up to model predictive control with CVXPi, as well as offering Python programming courses for electrical and mechatronic engineering. Want to master the entire series, from vehicle models to circuit testing on winding paths with various speeds? All 6 episodes, complete with PDF step-by-step guides and source code, are available to premium tutorial subscribers. Subscribe starting at Rp69.000 per month, save 28% with the 12-month package, or pay once Rp999.000 for lifetime access. Want just one class? Buy individual classes with lifetime access. Subscribe now on the premium page cerdasinovasidigital.com Contact our WhatsApp or visit cerdasinovasidigital.com Thank you for watching, see you in the next episode.
Ask the class material
Type a question or keyword; we find the part of this episode that explains it and jump the video there.
My notes
Notes are saved with the current video time and are only visible to you.
Questions & answers
Stuck on a step? Ask here; the instructor and other learners can answer.
No questions yet.
Your instructor
Cerdas Inovasi Digital Team
Software developers, BIM & 3D engineers, and AI practitioners
53 video classes1,283 episodes
We build software, BIM models, games and AI automations for clients every day, and turn that hands-on work into step-by-step classes: every step is shown on screen and comes with a PDF guide you can keep.
