Autonomous vehicles learn at home during lockdown

Autonomous vehicles learn at home during lockdown

As the world is forced to stay at home for an unprecedented duration owing to the COVID-19 pandemic, firms which were working on self-driving automobiles have been badly hit. For them, being off road means precious data-gathering time lost forever – because the more on-road data they can gather, more accurately can their automation systems be refined.

The autonomous vehicle industry is yet at a nascent stage where they need to constantly gather on-the-road situational data under diverse driving conditions – both normal and unusual. Readings from these real-life cruises are fed into the AI for further analysis – based on which the system keeps on refining itself – both in terms of functionality and safety. To ensure secure test-drives, each test-cruise requires two navigators – but social distancing norms currently cannot allow that. Not getting on road is simply fallow time for these firms, cutting into both their bottom-lines as well as projected deadlines.

Faced with an indefinite lockdown, such companies are now desperately seeking ways to keep their heads above water.While San Francisco-based self-driving trucking startup Embark Truckshave closed their workshop till the situation improves – which nobody knows when – layoffs have started to rear its ugly head at other self-driving car developers like Ike, Zoox, LyftandKodiak Robotics.

However, some smart companies are utilizing this hiatus as an opportunity to review and refine the algorithm and data models they use in their AI-based driving systems. Most of them still hold a backlog of un-utilized data that are now being analyzed, labelled in detail, used in 3D mapping – all ploughed back to fortify the existing loopholes in the system.

A recent article in MIT Technology Review lists some of the ways in which such companies are keeping their staff on the mark.

  • Vehicle operators at Aurora Innovation have been assisting their triage and labeling teams in data mining activities. According to their CEO Chris Urmson, this “gives them better context into our overall development process and will help them be even better at their job as we get back on the road.”
  • Embark Trucks is currently focusing on software to test their hardware components offline.
  • General Motors-owned Cruise is using advanced simulators to keep their autonomous software up and running – all the while improving on details to better assess competency in unusual situations, like how self-driving vehicles should react to ambulances on road.

Although such back-end activity is not apparently fruitful in the short term, almost all of the companies admit this will reap rich rewards in the long run.  And when everyone returns on road, those who had done their homework right during this downtime will outperform the others.

Leave a comment

Your email address will not be published. Required fields are marked *

© 2023 Praxis. All rights reserved. | Privacy Policy
   Contact Us
Praxis Tech School
PGP in Data Science