How to develop automotive grade applications

Hi all,
I get asked a lot about how do we approach the development of automotive grade applications. Our own Chris has written an excellent blog post about how to design a LiDAR-based (specifically NDT-based) localization algorithm:

The article is 10min long but entertaining and hopefully explains that writing requirements, doing failure analyses or developing test-driven code actually results in fantastically reliable code.

This is part 1 and we will follow with the part 2 (implementation) and part 3 (testing) in the coming weeks.

LMK what you think, D.


This is a fantastic article, and I’m looking forward to reading parts two and three. The explanations are clear and it lays out the steps to follow well. I’d like it to be required reading for the engineers here.


Hi all,
we published the part 2 (of 3) of building the safe and robust algorithms for autonomous driving in the open This part is about the implementation and we implemented the NDT and the MPC algorithms.

Video of the NDT matching:

Part 1 (design) was published 3 months ago:


There is a much better video for the illustration of an “aha” moment than the one used. :stuck_out_tongue:

1 Like

Hi all, we published the part 3 of 3 of the series called Building Safe Algorithms in the Open:

For refresher these were the first 2 parts:


These 3 articles explore how to implement 2 well known families of algorithms (Model Predictive Control and Normalized Distributed Transform) in the open source with the implementations being robust and safe.