Achievement Unlocked! With three companies now having launched driver-out deployments in the United States (Waymo, Cruise, and May Mobility), we've shown that the autonomous driving technology can work. Achievement unlocked! The next milestone is making the business work. Getting vehicles to operate with billions of dollars of engineering effort, expensive sensors + compute is one thing-- getting vehicles to operate where the revenues justify the R&D costs and vehicle costs is quite another. Even if the vehicles are amazing, recouping billions of dollars of R&D investment is a challenge. Emerging AI technologies are showing that similar-- if not better-- performance can be achieved with far lower R&D costs. We're not the only ones investing in these emerging methods, but we're the only ones that generate new training data on-the-fly and learn in-situ, structurally addressing the "edge case" problem that vexes virtually all data-driven approaches. (That includes end-to-end learning methods.) Low R&D and per-vehicle costs make it possible to not only develop the technology but also build a business case around it. We're live in six cities, generating positive margins, allowing us to grow our business in a capital efficient way. We expect to become the first profitable AV company, and in the process, be able to scale faster than anyone else.
How do you do safety assurance for these development approaches?
May Mobility can do it
There is no AI technology that can be trusted to make safety critical decisions at this time. AI does not know how to interpret edge cases it has never seen and make good safety decisions.
Kirk Steudle
3w
Great work Ed. It’s fascinating to watch your growth and tenacity to stay focused on your original vision. Congratulations.
Ryan Morton
3w
Keep it up Ed & May Mobility team!
Rick LeBlanc
3w
Excellent
D Greene
3w
Good Luck!
Well done!
Congrats Edwin!
Director of the QUT Centre for Robotics, ARC Laureate Fellow, ATSE Fellow. Positioning Systems for Robots and Autonomous Vehicles. Expert Speaker & Advisor on Autonomous Vehicles, AI & Robotics. Educational entrepreneur.
3wThanks Edwin Olson for bring some grounded reality to the AV discussion - you talk a lot about R&D costs, which is understandable, but in the medium term, isn't one of the primary issues the ongoings (e.g. control centres etc. etc. etc.) which from many conversations I've learnt are higher than someone might have expected especially in thin margin scenarios - can you comment on that, or do you think that's already within realms of plausible profitability?