In the previous post, I made a discussion about the challenges of self-driving tech development and its difficulty to achieve the goal. Two months ago, a report in McKinsey surveyed 75 automakers/transportation/self-driving startup companies, making prediction for the timeline of fully self-driving technology.
In general, there are three big challenges made level 4/5 deployment hard to continue: law
In McKinsey’s self-driving tech prediction report, they classify the technology into 6 types…
Level 4 is considered to be fully autonomous driving. Tech in this level is developed to on-street parking, parking in garages, highway pilot, L4/L5 Robotaxi, plus truck delivery “hub to hub” thru highway, and full journey on/off highway.
The one which could affect everyone's life, also the highest economic value part, is Robotaxi. This service is expected to be on the road in 2028. Two years ago I predicted the service will start in 2027. After several years’ observation, I fix my prediction to 2030. My opinion might be more conservative than those car companies’ policymakers, but sometimes our visions are quite objective.
Overall speaking, companies and startups from USA and Asia, they thought new techs might start earlier. But European and traditional automakers are not so optimistic. I guess readers would not be surprised about this result.
While discussing about fully self-driving technology timeline, besides traffic regulations, auto insurance regulations, technical ethics regulations, cyber security controls, route planning, parking lot planning, driver license regulations and professional drivers’ right to work, I think the basic tech threshold is MPI (Miles Per Intervention).
Based on US government data, human-error car accident happened once in every 200,000 miles. The leading company of the CA road test last year like AutoX, however, they only reach 50,000 miles. Looking back the road test reports in last a few years, and also considering software and hardware developing progress, these tech companies still need 5 to 7 years to reach 200,000 miles landmark.
Having said that, I still suggest that companies should set “safety index” at 2 ( in another words, 400,000 miles), for the high safety standard sake, to have drivers use artificial intelligence (AI) with high confidence to travel between destinations without a human operator.
In this case, even taking Moore's Law ( integrated circuit (IC) performance doubles about every 18 months) into consideration, timeline will be beyond 2030 for companies to improve “safety index” to 2.