Brad Templeton

The Waymo robotaxi service has been having a bad week, particularly in dealing with flooded roads. Waymos have driven into flooded streets and gotten stuck.

Through a mix of luck and skill, I happen to have made the most accurate prediction of the timelines for robotaxis back in 2013.

As a robocar team progresses towards its eventual goal of making a fully unsupervised vehicle that can drive without the full-time involvement of a human, they will go through many steps.

Waymo's new study refines autonomous vehicle safety comparisons by factoring in incident rates specific to location, road, and time, aiming for a true "apples-to-apples" evaluation against human driving. This approach provides crucial data for regulators to assess if robocars genuinely improve overall road safety, contrasting with past misleading claims like Tesla's exaggerated 10x safety improvement. The article stresses that robocars will always make mistakes, and public focus on individual incidents often overshadows the larger goal of reducing serious injuries. Regulators should prioritize data showing improving mistake rates and potential for significant risk reduction, even tolerating minor inconveniences. The focus should be on public interest goals, not prescriptive technical standards.

Two recent proposed regulations for robotaxis have come forward in DC and New Jersey, each supported by a different robotaxi player.

A new self-driving transit concept, "Vansit," aims to revolutionize urban mobility by offering a superior alternative to private cars. Utilizing 9-16 seat autonomous vans, Vansit detects groups sharing travel routes. Riders are transported by individual robotaxis or bikes to a merge point, where a waiting self-driving van takes them non-stop along the common route in luxury. At divergence points, other autonomous vehicles complete their journey. This system promises to be more efficient, cost-effective (1/5th the price of a car), and faster, significantly reducing traffic and energy consumption. Crucially, it leverages self-driving for seamless first/last mile connections, ensuring riders never wait, making it a truly "better-than-car" experience for 21st-century transportation.

Summary The San Francisco Bay Area's BART system is planning a costly $12.7 billion, 6-mile extension to San Jose, slated for completion by 2037, funded partly by local sales taxes. This project proceeds despite BART facing a severe post-COVID ridership crisis, threatening station closures and highlighting its expensive, non-standard design and high operating costs. The article contrasts this with 21st-century transit alternatives, such as systems using smaller, lighter vehicles, which promise significantly cheaper construction and operation. These modern solutions offer more efficient, non-stop service with flexible, low-cost stations, potentially costing $50 million per mile compared to BART's $2 billion per mile for tunnels. While radical, BART's financial struggles might eventually compel consideration of these superior, cost-effective transportation options.

Analyzing NHTSA robocar crash data by estimating fault provides critical insights into autonomous vehicle safety, a departure from traditional research. This method is vital for assessing robocars aiming for superhuman driving. Waymo exhibits a strong safety profile, being at fault in only 12-15% of its many reported incidents, frequently getting hit while stopped, which also indicates extensive operational mileage. Conversely, Tesla's fewer reported crashes show a higher at-fault rate, even under supervision, suggesting less real-world driving and potential system immaturity. The number of 'not-at-fault' incidents serves as a key indicator of actual miles driven. To ensure unbiased evaluations, the article stresses the need for independent fault assessments, leveraging robocars' detailed 3D recordings, rather than company-reported narratives.

EV charging is poised for a major transformation as self-driving technology advances. While current stations often mimic gas stations, a significant shift is coming: most EVs will soon navigate themselves to charging points. The key challenge remains automated plugging. The article proposes a return to "full-serve" stations, where human attendants or future robotic arms handle the connection. This would create a "no experience at all" scenario for owners, as cars autonomously seek out charging, potentially optimizing for cheaper electricity during off-peak hours or solar surpluses. This model solves issues for those without home charging, reduces grid strain, and allows stations to be located in less prime, more cost-effective areas, offering additional car services like washes.

A Waymo robotaxi was involved in a fatal Dallas crash where a pedestrian, struck by another SUV while jaywalking, was thrown into the Waymo's path. The pedestrian was hit by an SUV traveling 35mph, then flung 30 feet into the oncoming lane. The unoccupied Waymo, moving at 24mph, detected the pedestrian 150 feet away and slowed, impacting the pedestrian at 5mph. Police and Waymo reports suggest the robotaxi was not at fault, and the low-speed contact was unlikely the primary cause of death. This incident draws parallels to the Cruise crash, but Waymo has been transparent, unlike Cruise, whose poor post-crash communication led to its shutdown. The article questions if Waymo's "superhuman" capabilities could have prevented or further mitigated the crash, even if not legally liable. Given the suddenness, it's concluded Waymo likely couldn't have reliably prevented contact, though such incidents offer learning opportunities for autonomous systems.