US Lawmakers have recently passed legislation that allows fully autonomous vehicles to share public roads. With their potential to revolutionize the transport experience — and to improve road safety and traffic efficiency — there is a strong push by vehicle manufacturers and government agencies to bring autonomous to the broad market. The recent demonstrations at the DARPA Grand Challenges and by industry leaders has established that the core technical barrier to achieving autonomous vehicles is road scene understanding. However, although vehicle infrastructure, signage, and rules of the road have been designed to be interpreted fully by visual inspection, the use of computer vision in current autonomous vehicles is minimal. There is a perception that a wide gap exists between what is needed by the automotive industry to successfully deploy camera-based autonomous vehicles and what is currently possible using computer vision techniques. The goal of this workshop is to bring together leaders from both academia and industry to determine the true extent of this gap, to identify the most relevant aspects of computer vision problems to solve, and to learn from others about proposed avenues and solutions. Within the scope of the workshop will be core computer vision tasks such as dynamic 3D reconstruction, pedestrian and vehicle detection, and predictive scene understanding — all required capabilities for an autonomous vehicle. In particular, we will cover (but not limit ourselves to) the following questions in this workshop: Are current methods and representations adequate to hand over the wheel to computer vision algorithms? Are current benchmarks and datasets sufficient to support the on-going research? What are the mission-critical problems that need to be addressed by priority?