Abstract: Described herein are systems and methods for determining a trajectory for a vehicle. In one example, a system includes a processor and a memory in communication with the processor having a planning module. The planning module includes instructions that, when executed by the processor, cause the processor to determine, using a unified neural network based on input information, ego vehicle future trajectories of the ego vehicle and agent future trajectories of one or more agents, select one of the ego vehicle future trajectories as a selected trajectory based on the ego vehicle future trajectories and agent future trajectories using a cost function, and cause the ego vehicle to execute the selected trajectory.

Abstract: Systems and methods for training a policy are disclosed. In one example, a system includes a processor and a memory with instructions that cause the processor to train the policy using a training data set with training scenes to generate an identification policy and perform a closed-loop simulation on the identification policy to collect closed-loop metrics. Based on the closed-loop metrics, the instructions cause the processor to construct an error set of the training scenes and construct an upsampled training set by upsampling the error set. After that, the policy is trained using the upsampled training set to generate a final policy.

Abstract: In one embodiment, a method includes accessing sensor data associated with an environment of the vehicle; identifying agents in the environment based on the sensor data; determining a probability distribution indicative of whether a preliminary trajectory of each of the agents intersects a potential travel space of the vehicle; generating a prioritization of the agents based on the probability distribution; allocating an amount of computing resources of the vehicle for analyzing each of one or more of the agents based on the prioritization and characteristics of the respective one or more agents; and predicting a trajectory of one or more of the agents according to the allocated computing resources. An accuracy of the prediction is proportional to the amount of allocated computing resources. The method also includes determining, based on the analysis of the agents, one or more operations for the vehicle to perform.

Abstract: A method includes accessing perception data generated based on first sensor data and generating first prediction data using a prediction module based on the perception data. The method includes capturing second sensor data while the vehicle is operating according to a first planned trajectory based on the first prediction data and a path planning parameter. The method includes generating a simulation for evaluating the prediction module, including generating second prediction data based on the second sensor data and generating a second planned trajectory. The method includes, subsequent to determining that a difference between the first planned trajectory and the second planned trajectory fails to satisfy predetermined prediction criteria, identifying an object for which the prediction module underperforms relative to the predetermined prediction criteria and updating the prediction module based on the identified object and data associated with the second prediction data.

Abstract: Systems, methods, and non-transitory computer-readable media can detect an occurrence of a condition in an environment based on sensor data captured by a vehicle. A determination is made whether the occurrence of the condition satisfies a threshold associated with a likelihood that a behavior associated with an object in the environment will occur based on an interaction between the condition and the object, wherein the likelihood is based on prior observations of one or more objects. Subsequent to determining that the threshold is satisfied, a vehicle operation that is associated with the likelihood that the behavior associated with the object will occur is performed.

Abstract: Systems, methods, and non-transitory computer-readable media can receive a context comprising attributes associated with an agent. The attributes associated with the agent can be determined based at least in part on sensor data captured by at least one vehicle. A plurality of trajectory prediction algorithms to be applied to the agent can be evaluated to generate a plurality of performance metrics associated with the plurality of trajectory prediction algorithms. A particular trajectory prediction algorithm of the plurality of trajectory prediction algorithms that is to be applied to the agent can be determined based on evaluating the context against the plurality of performance metrics. The context can be associated with a particular cluster of a plurality of clusters of contexts such that the particular trajectory prediction algorithm is associated with the particular cluster.

Abstract: In one embodiment, a method includes accessing sensor data associated with an environment of the vehicle; identifying agents in the environment based on the sensor data; determining a probability distribution indicative of whether a preliminary trajectory of each of the agents intersects a potential travel space of the vehicle; generating a prioritization of the agents based on the probability distribution; allocating an amount of computing resources of the vehicle for analyzing each of one or more of the agents based on the prioritization and characteristics of the respective one or more agents; and predicting a trajectory of one or more of the agents according to the allocated computing resources. An accuracy of the prediction is proportional to the amount of allocated computing resources. The method also includes determining, based on the analysis of the agents, one or more operations for the vehicle to perform.