Simulation has the potential to massively scale evaluation of self-driving systems enabling rapid development as well as safe deployment. To close the gap between simulation and the real world, we need to simulate realistic multi-agent behaviors. Existing simulation environments rely on heuristic-based models that directly encode traffic rules, which cannot capture irregular maneuvers (e.g., nudging, U-turns) and complex interactions (e.g., yielding, merging). In contrast, we leverage real-world data to learn directly from human demonstration and thus capture a more diverse set of actor behaviors. To this end, we propose TrafficSim, a multi-agent behavior model for realistic traffic simulation. In particular, we leverage an implicit latent variable model to parameterize a joint actor policy that generates socially-consistent plans for all actors in the scene jointly. To learn a robust policy amenable for long horizon simulation, we unroll the policy in training and optimize through the fully differentiable simulation across time. Our learning objective incorporates both human demonstrations as well as common sense. We show TrafficSim generates significantly more realistic and diverse traffic scenarios as compared to a diverse set of baselines. Notably, we can exploit trajectories generated by TrafficSim as effective data augmentation for training better motion planner.
To date, some of the hottest players in the autonomous arena have put their energy into driving endless miles in order to fine-tune their AI-based technology. It's time consuming and expensive, and most approved test driving sites have been in parts of the world with sunny skies and dry roads, hardly indicative of common driving conditions. With this mile-by-mile method, it's hard to imagine self-driving could become commonplace anytime soon. However, there is another approach to training and tuning AI for optimized self-driving - cutting-edge simulators that enhance deep learning and offer any number of driving scenarios to train the AI, helping to bring autonomous vehicles (AVs) to market in a more reasonable timeline. The newest approaches to self-driving recognize that by relying on a simulated testing environment, development gets an enormous boost in terms of cost, time, and safety.
Sitting in traffic is miserable. It's bad enough when the car congestion is caused by construction or crashes, but it's a far worse feeling to get through a stop and go traffic blockage and find that there was no discernible reason for the slowdown. Why do these slow spots pop up along the highways? That's what computer models like Traffic-Simulation are designed to figure out. These simulations model the effects of changing various traffic conditions--such as the number of cars and trucks on the road, to the average distance between cars, average speed, and other factors--to see when and how traffic jams develop.
This article gives an introduction to agent-based modeling and simulation (ABMS). After a general discussion about modeling and simulation, we address the basic concept of ABMS, focusing on its generative and bottom-up nature, its advantages as well as its pitfalls. The subsequent part of the article deals with application-oriented aspects, including selected tools and well-known applications. In order to illustrate the benefits of using ABMS, we focus on several aspects of a well-known area related to simulation of complex systems, namely traffic. At the end, a brief look into future challenges is given.
Traffic evacuation plays a critical role in saving lives in devastating disasters such as hurricanes, wildfires, floods, earthquakes, etc. An ability to evaluate evacuation plans in advance for these rare events, including identifying traffic flow bottlenecks, improving traffic management policies, and understanding the robustness of the traffic management policy are critical for emergency management. Given the rareness of such events and the corresponding lack of real data, traffic simulation provides a flexible and versatile approach for such scenarios, and furthermore allows dynamic interaction with the simulated evacuation. In this paper, we build a traffic simulation pipeline to explore the above problems, covering many aspects of evacuation, including map creation, demand generation, vehicle behavior, bottleneck identification, traffic management policy improvement, and results analysis. We apply the pipeline to two case studies in California. The first is Paradise, which was destroyed by a large wildfire in 2018 and experienced catastrophic traffic jams during the evacuation. The second is Mill Valley, which has high risk of wildfire and potential traffic issues since the city is situated in a narrow valley.