Automated creation of synthetic traffic scenarios is a key part of scaling the safety validation of autonomous vehicles (AVs). In this paper, we propose Scenario Diffusion, a novel diffusion-based architecture for generating traffic scenarios that enables controllable scenario generation. We combine latent diffusion, object detection and trajectory regression to generate distributions of synthetic agent poses, orientations and trajectories simultaneously. This distribution is conditioned on the map and sets of tokens describing the desired scenario to provide additional control over the generated scenario. We show that our approach has sufficient expressive capacity to model diverse traffic patterns and generalizes to different geographical regions.

There has been a considerable amount of recent work on high-speed micro-aerial vehicle flight in unknown and unstructured environments. Generally these approaches either use active sensing or fly slowly enough to ensure a safe braking distance with the relatively short sensing range of passive sensors. The former generally requires carrying large and heavy LIDARs and the latter only allows flight far away from the dynamic limits of the vehicle. One of the significant challenges for high-speed flight is the computational demand of trajectory planning at sufficiently high rates and length scales required in outdoor environments. We tackle both problems in this work by leveraging semantic information derived from an RGB camera on-board the vehicle. We first describe how to use semantic information to increase the effective range of perception on certain environment classes. Second, we present a sparse representation of the environment that is sufficiently lightweight for long distance path planning. We show how our approach outperforms more traditional metric planners which seek the shortest path, demonstrate the semantic planner's capabilities in a set of simulated and excessive real-world autonomous quadrotor flights in an urban environment.

There has been considerable recent work in motion planning for UAVs to enable aggressive, highly dynamic flight in known environments with motion capture systems. However, these existing planners have not been shown to enable the same kind of flight in unknown, outdoor environments. In this paper we present a receding horizon planning architecture that enables the fast replanning necessary for reactive obstacle avoidance by combining three techniques. First, we show how previous work in computationally efficient, closed-form trajectory generation method can be coupled with spatial partitioning data structures to reason about the geometry of the environment in real-time. Second, we show how to maintain safety margins during fast flight in unknown environments by planning velocities according to obstacle density. Third, our receding-horizon, sampling-based motion planner uses minimum-jerk trajectories and closed-loop tracking to enable smooth, robust, high-speed flight with the low angular rates necessary for accurate visual-inertial navigation. We compare against two state-of-the-art, reactive motion planners in simulation and benchmark solution quality against an offline global planner. Finally, we demonstrate our planner over 80 flights with a combined distance of 22km of autonomous quadrotor flights in an urban environment at speeds up to 9.4ms $^{-1}$.