Conference Presentation
Emilio Frazzoli, CTO, nuTonomy - MIT Self-Driving Cars
- Full autonomy (Level 4 or 5) is deemed essential for capturing technology value, with lower levels failing to provide necessary convenience or enable effective car sharing.
- Recovered driving time is projected to yield $1.2 trillion annually in economic value, surpassing the value of increased safety, while a functional, shared autonomous vehicle system could generate approximately $2 trillion per year by substituting one shared vehicle for four privately owned ones.
- Autonomous vehicles will become common as a summonable service within a couple of years, whereas private ownership is not expected for at least 20 years.
- Cost constraints differ by model: private buyers require autonomy packages priced at roughly $20,000 or less based on a 10-year recovery of driver time value, while service models face no such constraint given human driver costs of approximately $100,000 annually.
- Safety improvements are anticipated to be asymptotic, with uncertainty regarding when fully autonomous vehicles will consistently outperform human drivers.
- High-definition maps are expected to become ubiquitous within a few years as fleet operators generate mapping data autonomously through daily operations.
- The industry faces divergent paths, with OEMs pursuing a Level 0-5 progression while entities like Waymo and Uber target full automation starting from small geofenced areas.
- Significant challenges include the lack of a rigorous theory for vehicle behavior due to inconsistent road rules, and the need for community agreement on ethical decision-making parameters.
- Market shifts driven by increased autonomous mobility supply are predicted to lower mobility costs and reduce wages for human drivers.
- Neutonomy intends to double its size within the next couple of years by hiring a few hundred employees.