Planning for Vehicle Automation Jeremy Raw, P.E. Office of Planning Federal Highway Administration October 27, 2016 AMPO Annual Meeting, Fort Worth, TX October 27, 2016 Planning for Vehicle Automation

Planning for Vehicle Automation Overview Automation is a Disruptive Technology Performance-Based Planning to the Rescue New Roles for Scenarios and Models A “big deal”, but not in the way you might immediately think “Disruptive” of the planning process, that is A good process beats bad assumptions every time It would be mean to send you home without some practical suggestions October 27, 2016 Planning for Vehicle Automation

Disruptive Technology “Disruptive” = threat to “business as usual” Automation could upend how transportation works Most disruptive elements: How or when will it happen? What will the net effects be? What to do now (or next)? “Disruptive” means the end of “business as usual”. But what business are we in if it’s not usual? – we’ll return to that. October 27, 2016 Planning for Vehicle Automation

What makes Automation Disruptive? Behavior Drive farther, buy fewer cars, use more taxis Infrastructure Better/different markings, less new road capacity, fewer parking spaces, “internet of cars” Agency roles Transit vs. shared mobility, “hands on” system management & operations A short litany of things that AVs are anticipated to affect. October 27, 2016 Planning for Vehicle Automation

What Else is Disruptive? Self-driving cars are not the only thing Shared Mobility (Uber/Lyft) Freight Management (self-driving trucks, drones, 3D printing) Bicycle and pedestrian travel (“Return to the City”) Generational shifts (“Millenials hate driving”) Externalities (new energy sources, climate change) Common point: we don’t know whether any of these will occur or continue, or what the effects will be if they do. October 27, 2016 Planning for Vehicle Automation

Deployment Status: Don’t Panic Widespread AV deployment may be 30 years away Automated Vehicles defined by hype and hope. We’re still all the way down at the left of the technology deployment curve. There is no objective basis (yet) for determining how these technologies will perform in the real world, or what people will choose to do with them Deployment curve example: Insurance Institute for Highway Safety; ESC has been required by NHTSA in 100% of new cars since 2012 (55% in 2009, 75% in 2010, 95% of 2011) October 27, 2016 Planning for Vehicle Automation

Planning for Vehicle Automation The Future… … is Wide Open More opportunity than ever to improve … is Radically Uncertain Know less than ever about what will happen … calls for New Planning Strategies Extrapolate less, experiment more, get more data The rest of my talk is about the Planning Strategies October 27, 2016 Planning for Vehicle Automation

Performance-Based Planning to the Rescue October 27, 2016 Planning for Vehicle Automation

Handling the Unknown with Performance-Based Planning Scenario Development and Visioning Performance Measure Development Performance Targets Data Collection Try Things and Evaluate Them Revisit Assumptions Regularly October 27, 2016 Planning for Vehicle Automation

Planning for Vehicle Automation Where do we want to go? Exit the era of “Manifest Destiny” Certainty about interventions and outcomes Consensus vision of what matters Enter the era of “New Horizons” Uncertainty about interventions and outcomes Diverse visions of what might be possible From “Determinism” to “Free Will” October 27, 2016 Planning for Vehicle Automation

Planning for Vehicle Automation Data Gathering More Helpful Models More Helpful This diagram is about how to respond to the things we don’t know. Blue line is the “sweet spot” where we’re making the best type of response to varying amounts of available information. October 27, 2016 Planning for Vehicle Automation

Planning for Vehicle Automation Develop New Knowledge Data Gathering More Helpful Models More Helpful Explore New Phenomena The gray ovals indicate how we can effectively use models when we are “off the line”. In these cases, models support the development of our understanding, not prediction of the future. Side note if time: “how much the future is like the past” is conditioned by three things: (1) the nature of the phenomenon itself (ontology); (2) the state of our knowledge of the phenomenon (epistemology); and (3) how clear we are about the outcomes we want (teleology). October 27, 2016 Planning for Vehicle Automation

Rethinking the Role of Scenarios Old: Scenarios help decide what will happen Identify factors everyone can agree on Compute outcomes based on known relationships New: Scenarios help decide what we want Identify outcomes we hope for (or fear) Identify indicators of where we’re headed We often think of scenarios as helping us develop the right foundation for predicting the future. Ithe October 27, 2016 Planning for Vehicle Automation

Rethinking the Role of Models Old: Models are predictive Visualize outcome of well-understood interventions New: Models are exploratory Explore possibilities of hypothetical scenarios Obsolete: Cannot use models to evaluate outcomes October 27, 2016 Planning for Vehicle Automation

Imagine the Possibilities Use planning cycles (TIP / LRP) to review Scenario Planning, early and often Focus on “where are we do we want to go?” Use models sparingly to explore alternatives Focus on Data Collection and Interpretation What are the performance measures & targets? Don’t neglect Visioning Now more than ever, the future is ours to define “Support the good, impede the bad” October 27, 2016 Planning for Vehicle Automation