After the AV Hype – Implication for Investors and Startups

by Drue Freeman

Originally Published in Wevolver March 10, 2020

The autonomous vehicle (AV) hype is waning. Startup valuations are becoming more reasonable. Capital is tightening. This is not necessarily bad. It just reflects the realization that widespread deployment of fully autonomous (Level 5) vehicles is still at least a decade away.

The 2019 Connected Vehicle and Smart Mobility Hype Cycle. Image: Garnter

Technically, the basic problems of Sensing, Thinking/Learning, and Acting have been addressed. Under good weather and lighting conditions in environments with relatively few pedestrians, AVs perform extremely well. State-of-the-art perception systems (Cameras, Radar, and LiDAR) can detect most hazards better than humans. Artificial intelligence fuses these sensor inputs to recognize objects, precisely localize the vehicle in space, and plan the optimal path to reach the desired destination.

Unfortunately, these systems remain too expensive, power consuming, and data intensive to be suitable for mass deployment. The corner cases involving bad weather, poor infrastructure, and chaotic road conditions are proving to be tremendously challenging. Significant improvements are still required in the efficacy and cost efficiency of the existing sensors. New sensors, like thermal, will be needed which have the ability to see at night and in inclement weather. Similarly, AI computing must become more efficient as measured by meaningful operations (e.g., frames or inferences) per watt or per dollar. Furthermore, automotive grade networking protocols are not sufficient to move the terabytes of raw sensor data around the car in real time that these systems will be generating.

The good news is that hundreds of companies, from startups to established industry leaders, are still investing heavily in the required improvements, though not with the same wild abandon as they did just a few years ago. The problems will be solved, as will the necessary regulatory challenges hindering widespread adoption of AVs. But the realization that broad Level 5 deployment will take longer than previously hoped means investors must place their bets more carefully and be a little more patient for a return. In the meantime, Level 4 deployments in trucking, mining, agriculture, and delivery – use cases in which reasonable geofencing can be put in place and for which cost efficiencies may not be as critical – will help continue to advance the technology.