The New Evidence: More Automation, More Miles, More Congestion
For years, visions of the future of urban mobility have featured self-driving cars gliding through calm, free-flowing streets. New research suggests the reality of self driving traffic could be very different. A meta-analysis from the University of Texas at Arlington examined how autonomous vehicles affect vehicle miles traveled (VMT) and found an average 5.95% increase in VMT, rising to nearly 7% for non-shared autonomous vehicles. That may sound modest, but even small percentage jumps compound quickly on already overloaded roads. The mechanisms are straightforward: easier, more comfortable trips encourage people to travel farther and more often; cars can circle or reposition themselves without passengers; and pickup and drop-off activity multiplies. In other words, autonomous vehicle congestion is not a paradox—it is a predictable consequence of making car travel cheaper in time and effort, unless policy and design intervene.
Robotaxis on the Street: Smooth Rides, Real‑World Snarls
Early deployments reveal how the self driving car impact is playing out in real traffic. In Tokyo, Nissan’s Ariya prototype navigated 15km of dense downtown driving “like a competent, considerate human,” handling bridges, pedestrians, and complex streets without driver intervention. Engineers even argue its 360-degree attention and training from millions of driving episodes make it technically equal to or better than a human driver. Yet in robotaxi pilot cities, the picture is less serene. Waymo vehicles have frozen at intersections during network problems, including dozens stuck during a power outage that clogged streets for hours. In another incident, Baidu robotaxis simultaneously stopped on elevated highways, trapping passengers mid-traffic. These episodes are happening while fleets are still limited. Multiply robotaxi empty trips and occasional failures across thousands of vehicles, and small glitches start to create system-wide traffic shocks instead of relief.
Why Self‑Driving Can Jam Roads: Empty Trips and Mode Shifts
The biggest congestion threat is not that autonomous cars drive badly—it is that they drive too much. When a car can chauffeur itself with no one inside, new travel behaviors emerge. Commuters might send their vehicle home after drop-off, instruct it to circle the block to avoid parking costs, or dispatch it across town for errands they once did on foot. Every such decision adds zero-occupant miles. Robotaxi empty trips between passengers also inflate VMT, particularly if vehicles deadhead long distances in search of riders. At the same time, more convenient door-to-door service can lure riders away from buses and trains, reducing public transit’s share and putting more vehicles on the road. Without constraints, autonomous vehicle congestion becomes a textbook example of induced demand: when driving becomes easier and more comfortable, people do more of it—even if it slows everyone down.
Deployment Models Matter: Private AVs vs Robotaxis vs Shared Shuttles
Not all self-driving traffic looks the same. Private autonomous cars are the worst-case scenario for congestion: they maximize comfort and flexibility while making zero-occupant trips trivial. The University of Texas at Arlington study shows non-shared AVs push VMT increases toward 7%, largely because each car serves only one household yet can roam freely. Robotaxis can be better, but only if they are highly utilized and carefully managed. Poorly routed fleets may rack up robotaxi empty trips that offset any gains from ride-pooling. Shared autonomous shuttles and fixed-route services are the most promising for reducing congestion, because they concentrate demand into fewer vehicles and can complement mass transit. Ultimately, the self driving car impact on traffic depends less on algorithms behind the wheel and more on the service model: private luxury pods or shared, transit-like mobility.
Avoiding a Shinier Traffic Jam: Policy, Pricing and Realistic Expectations
Autonomous vehicles are not destined to ruin or rescue our roads; governance will decide their role in the future of urban mobility. Policymakers can steer outcomes by pricing road use—such as congestion charges or fees for zero-occupant cruising—so that empty repositioning trips do not overwhelm streets. Zoning and curb-management rules can limit chaotic pickup and drop-off behavior, while fleet regulations can require minimum occupancy rates, data sharing, and strong integration with public transit. Cities can prioritize shared AV services over private ownership through parking policy and dedicated lanes, ensuring automation amplifies, rather than competes with, high-capacity transport. For the public, the myth that self-driving will automatically erase traffic needs updating. Over the next decade, automation may improve safety and comfort, but without firm policy guardrails, the most likely outcome is a more convenient, more automated, and just as congested commute.