Since the days of the DARPA grand challenge, Autonomous Vehicle (AV) Technology has generated a great deal of excitement. Companies such as Google
One of the most interesting areas for investment has been in the area of the area of test tracks. A sampling of recent announcements include:
- New Autonomous Vehicle Test Track to Open in Metro Atlanta
- South Fayette approves plan for autonomous vehicle test track
- Minister opens 70km test track ‘village’ to ensure autonomous vehicle safety (Millbrook, UK)
The typical situation is to take an older asset such as a closed airport and with some investment of some civil infrastructure, declare it an autonomous vehicle test track. Interestingly, until recently, automotive test tracks were actually in decline. In 1991, the New York Published an article “All About/Automobile Testing; The Test Track Is King Of the Road No More,” which explained the impact of simulation on the usage of traditional test tracks. Automotive OEMs preferred to use simulation because it was cheaper, safer, and allowed for more through testing. Test tracks were relegated to the tasks which could not be done well with simulation such as reliability or broad environmental testing.
Will Autonomous Vehicles change this secular decline in test tracks ?
Well, the “grand daddy” of test tracks is considered to be the American Center for Mobility (ACM). Founded in 2018, ACM is a massive 500 acre facility which was launched with great fanfare, strong commitments from the state of Michigan and an impressive set of partners. However, the sequence of subsequent events which followed was:
- Apr 2018: “The American Center for Mobility is open. What’s next?”
- Aug 2018: “American Center for Mobility parts ways with CEO; interim chief named”
- Feb 2019: “American Center for Mobility welcomes new CEO Michael Noblett”
- Nov 2019: “CEO of American Center for Mobility out after 9 months”
- Apr 2020: “American Center for Mobility names new CEO”
Five CEOs in 2+ years ! Is the issue the executives ? Very unlikely, but rather the real problem is very likely that the challenges of running an AV Test Track are exceedingly difficult. These challenges include:
- Technology: Today, most test tracks largely consist of the civil engineering shell (roads, turns, highway canyon, intersections, etc). The exact requirements for AV testing are not currently known because of the immaturity of AV technology. Open questions include: what is the required connection to simulation ? Likely one needs to build an infrastructure more like a movie set. What about testing sensors or communication ? One needs a lot of electronics equipment for that purpose. Thus, it is likely that there will be a need for much deeper investments, but who will make these investments ? What is the justification ?
- Business: Certainly there is business driven by Federal government research (Mcity or TRC as examples), but is it enough to sustain all of these AV tracks? In addition, most OEMs/large companies have their own private tracks. Will they need more capacity ? If the business is to come from startups, what additional engineering services will have to be provided to engage with them ? Also, what will be the impact of the deflation of the AV hype cycle? Perhaps most importantly, the end customer has not yet figured out how to built a sustainable business, so how can a supporting resource such as a test track ?
- Liability and Capability: The nature of safety testing is to test at the edge of safety. Thus, liability is a huge concern. For capability, transportation has expertise in the mechanical and civil engineering area. The requirement in the world of AV is for electrical engineering and computer science. Will test tracks be able to attract these resources in the context of a very competitive market for these skills.
- Expectations: Finally, publicly financed test tracks are typically built on the justification of improving safety and driving local economic development. Both are difficult to demonstrate.
Overall, a test track general manager is staring into a situation of extreme technical and business risk. Drumming up business (outside of federal grant resources) is very challenging.
Is there potentially a more lucrative business model for these assets ?
Tolling technology does not generate nearly the excitement of autonomous driving technology. However, perhaps in the short term it will have more impact. Tolling (or its dual crediting) is an extremely important technology because it provides an economic mechanism to manage the shared resources of the road network. As explained in “Could Micro-tolling Be The Answer To The Gas Tax And Traffic Optimization?,” without market based resolution of demand and supply, the road system suffers from oversupply (empty roads) and shortages (congestion). Average utilization of the road network is fairly low, so there is plenty of capacity available with some rescheduling. Finally, with the move to electric cars, the gas tax is no longer a reliable as an indicator for road usage. Further, mega trends such as ecommerce delivery, Mobility-As-A-Service (MAAS), grocery, and restaurant delivery create commercial fleets which enable further transaction based economic models.
Traditionally, limited access highways and parking have been the primary tolling structures. Recently, congestion based pricing has been introduced in large cities with area based tariffs. However, advanced tolling can enable a whole host of innovative mechanisms which more optimally use the transportation system for everyone’s benefit. The policy structures can be quite inventive (rewards points, park-and-ride, encouraging the use of pedestrian robots for shopping, etc), but underlying all of these methods is a basic technology which marks an individual at a time and place. This core capability can be rolled out in a variety of situations such as curbside time based access, dynamic urban canyon area tolling, and many more use-models.
How does one test these capabilities ? The civil infrastructure for an AV test track is an almost perfect resource. Looking at the four elements:
- Technology: Both camera and RF technology is reasonably well understood in this space, and testing the use models in various situations makes a great deal of sense. The civil infrastructures of urban, intersection, highway, campus, and others fit well for the application. Attaching this physical world to a tolling operating system to test the backend processing also makes a great deal of sense. Overall, the technology risk is quite manageable, and provides great leverage of the civil engineering assets.
- Business: Unlike the AV market, the tolling market is already billions of dollars in size. Further, the new delivery business models enable new models of monetization. Finally, the customers are in a closer neighborhood (other governmental authorities). Thus, the business risk is quite manageable.
- Liability and Capability: The explicit objective is not safety, so the liability issues are quite a bit less pronounced on the test track. Interestingly, the end result of tolling may well create safer roads. In terms of capability, tolling technology is in the neighborhood of the current civil engineering infrastructure, and not as big a stretch as AV safety testing.
- Expectations: Advanced tolling can increase resources for funding civil infrastructure and ease congestion. Both of these elements are easily measurable.
Does this mean shared AV test tracks will never be a viable use model for testing safety ?
The validation and testing methodology for AVs is still an area of active research. Currently, simulation is the leading candidate for doing the bulk of the testing. AV test tracks will certainly have a role, but the exact requirements are not clear. Using these AV test tracks for tolling could well provide a stepping stone for such a future capability.