Abstract: Systems and methods for path planning with latent state inference and spatial-temporal relationships are provided. A system includes an inference module, a policy module, a graphical representation module, and a planning module. The inference module receives sensor data associated with a plurality of agents. The inference module also maps the sensor data to a latent state distribution to identify latent states of the plurality of agents. The latent states identify agents of the plurality of agents as cooperative or aggressive. The policy module predicts future trajectories of the plurality of agents at a given time based on sensor data and the latent states of the plurality of agents. The graphical representation module generates a graphical representation based on the sensor data and a graphical representation neural network. The planning module generates a motion plan for the ego agent based on the predicted future trajectories and the graphical representation.

Abstract: Systems and methods for path planning with latent state inference and spatial-temporal relationships are provided. In one embodiment, a system includes an inference module, a policy module, a graphical representation module, and a planning module. The inference module receives sensor data associated with a plurality of agents. The inference module maps the sensor data to a latent state distribution to identify latent states of the plurality of agents. The latent states identify agents as cooperative or aggressive. The policy module predicts future trajectories of the plurality of agents at a given time based on sensor data and the latent states of the plurality of agents. The graphical representation module generates a graphical representation based on the sensor data and a graphical representation neural network. The planning module generates a motion plan for the ego agent based on the predicted future trajectories and the graphical representation.

Abstract: A system and method for multi-agent reinforcement learning in a multi-agent environment that include receiving data associated with the multi-agent environment in which an ego agent and a target agent are traveling. The system and method also include learning single agent policies that are respectively associated with the ego agent and the target agent based on the data associated with the multi-agent environment. The system and method additionally include learning a multi-agent policy as an interactive policy that enables the ego agent and the target agent to account for one another while traveling to respective goals within the multi-agent environment based on the single agent policies. The system and method further include implementing the multi-agent policy to control at least one of: the ego agent and the target agent to operate within the multi-agent environment.

Abstract: According to one aspect, systems and techniques for lane selection may include receiving a current state of an ego vehicle and a traffic participant vehicle, and a goal position, projecting the ego vehicle and the traffic participant vehicle onto a graph network, where nodes of the graph network may be indicative of discretized space within an operating environment, determining a current node for the ego vehicle within the graph network, and determining a subsequent node for the ego vehicle based on identifying adjacent nodes which may be adjacent to the current node, calculating travel times associated with each of the adjacent nodes, calculating step costs associated with each of the adjacent nodes, calculating heuristic costs associated with each of the adjacent nodes, and predicting a position of the traffic participant vehicle.

Abstract: According to one aspect, systems and techniques for lane selection may include receiving a current state of an ego vehicle and a traffic participant vehicle, and a goal position, projecting the ego vehicle and the traffic participant vehicle onto a graph network, where nodes of the graph network may be indicative of discretized space within an operating environment, determining a current node for the ego vehicle within the graph network, and determining a subsequent node for the ego vehicle based on identifying adjacent nodes which may be adjacent to the current node, calculating travel times associated with each of the adjacent nodes, calculating step costs associated with each of the adjacent nodes, calculating heuristic costs associated with each of the adjacent nodes, and predicting a position of the traffic participant vehicle.

Abstract: According to one aspect, cooperative multi-goal, multi-agent, multi-stage (CM3) reinforcement learning may include training a first agent using a first policy gradient and a first critic using a first loss function to learn goals in a single-agent environment using a Markov decision process, training a number of agents based on the first policy gradient and a second policy gradient and a second critic based on the first loss function and a second loss function to learn cooperation between the agents in a multi-agent environment using a Markov game to instantiate a second agent neural network, each of the agents instantiated with the first agent neural network in a pre-trained fashion, and generating a CM3 network policy based on the first agent neural network and the second agent neural network. The CM3 network policy may be implemented in a CM3 based autonomous vehicle to facilitate autonomous driving.

Abstract: A system and method for multi-agent reinforcement learning with periodic parameter sharing that include inputting at least one occupancy grid to a convolutional neural network (CNN) and at least one vehicle dynamic parameter into a first fully connected layer and concatenating outputs of the CNN and the first fully connected layer. The system and method also include providing Q value estimates for agent actions based on processing of the concatenated outputs and choosing at least one autonomous action to be executed by at least one of: an ego agent and a target agent. The system and method further include processing a multi-agent policy that accounts for operation of the ego agent and the target agent with respect to one another within a multi-agent environment based on the at least one autonomous action to be executed by at least one of: the ego agent and the target agent.

Abstract: A system and method for multi-agent reinforcement learning in a multi-agent environment that include receiving data associated with the multi-agent environment in which an ego agent and a target agent are traveling. The system and method also include learning single agent policies that are respectively associated with the ego agent and the target agent based on the data associated with the multi-agent environment. The system and method additionally include learning a multi-agent policy as an interactive policy that enables the ego agent and the target agent to account for one another while traveling to respective goals within the multi-agent environment based on the single agent policies. The system and method further include implementing the multi-agent policy to control at least one of: the ego agent and the target agent to operate within the multi-agent environment.

Abstract: A system and method for providing probabilistic-based lane-change decision making and motion planning that include receiving data associated with a roadway environment of an ego vehicle. The system and method also include performing gap analysis to determine at least one gap between neighboring vehicles that are traveling within the target lane to filter out an optimal merging entrance for the ego vehicle to merge into the target lane and determining a probability value associated with an intention of a driver of a following neighboring vehicle to yield to allow the ego vehicle to merge into the target lane. The system and method further include controlling the ego vehicle to autonomously continue traveling within the current lane or autonomously merge from current lane to the target lane based on at least one of: if the optimal merging entrance is filtered out and if the probability value indicates an intention of the driver to yield.

Abstract: A system and method for providing cooperation-aware lane change control in dense traffic that include receiving vehicle dynamic data associated with an ego vehicle and receiving environment data associated with a surrounding environment of the ego vehicle. The system and method also include utilizing a controller that includes an analyzer to analyze the vehicle dynamic data and a recurrent neural network to analyze the environment data. The system and method further include executing a heuristic algorithm that sequentially evaluates the future states of the ego vehicle and the predicted interactive motions of the surrounding vehicles to promote the cooperation-aware lane change control in the dense traffic.

Abstract: A system and method for multi-agent reinforcement learning in a multi-agent environment that include receiving data associated with the multi-agent environment in which an ego agent and a target agent are traveling and learning a single agent policy that is based on the data associated with the multi-agent environment and that accounts for operation of at least one of: the ego agent and the target agent individually. The system and method also include learning a multi-agent policy that accounts for operation of the ego agent and the target agent with respect to one another within the multi-agent environment. The system and method further include controlling at least one of: the ego agent and the target agent to operate within the multi-agent environment based on the multi-agent policy.

Abstract: A system and method for providing online multi-LiDAR dynamic occupancy mapping that include receiving LiDAR data from each of a plurality of LiDAR sensors. The system and method also include processing a region of interest grid to compute a static occupancy map of a surrounding environment of the ego vehicle and processing a dynamic occupancy map. The system and method further include controlling the ego vehicle to be operated based on the dynamic occupancy map.

Abstract: According to one aspect, a system for robust localization may include a scan accumulator, a scan matcher, a transform maintainer, and a location fuser. The scan accumulator may receive a set of sensor data from a set of sensors mounted on a vehicle. The scan accumulator may generate a sensor scan point cloud output by transforming the set of sensor data from each sensor frame to a corresponding vehicle frame and calculate a fitness score, a transformation probability, and a mean elevation angle used to determine a scan confidence for the sensor data. The transform maintainer may receive GPS data, the scan confidence, and the matched sensor scan point cloud output and map tile point cloud data from the scan matcher, and determine whether the GPS data or the matched sensor scan point cloud output and map tile point cloud data is utilized for a map-to-odometer transformation output.

Abstract: Systems and methods for generating instructions for a vehicle to navigate an unsignaled intersection are provided. The method may include: generating an expected return over a sequence of actions of the vehicle; determining an optimal policy by selecting an action with a maximum value for the vehicle; executing dynamic frame skipping to expedite learning a repeated action of the vehicle; prioritize an experience replay by utilizing an experience replay buffer to break correlations between sequential steps of the vehicle; generate a plurality of state-action representations based on at least one of the expected return, the optimal policy, the dynamic frame skipping, or the prioritized experience replay; generate the instructions for navigating the unsignaled intersection based on the plurality of state-action representations; and transmit the instructions for navigating the unsignaled intersection to the vehicle such that the vehicle executes the instructions to navigate the unsignaled intersection.

Abstract: According to one aspect, uncertainty prediction based deep learning may include receiving, using a memory, a trained neural network policy ? trained based on a first dataset in a first environment, implementing, via a controller, the trained neural network policy ? in a second environment by receiving an input and generating an output y, calculating an uncertainty array U[T] for a time window T, wherein the uncertainty array is indicative of a level of uncertainty associated with an output sample distribution of the output across the time window T based on a temporal divergence, an entropy H, a variational ratio VR, and a standard deviation SD of the output y, and executing, via the controller and one or more systems, an action based on the uncertainty array U[T], such as discontinuing use of the trained neural network policy ?.

Abstract: A system for generating a model-free reinforcement learning policy may include a processor, a memory, and a simulator. The simulator may be implemented via the processor and the memory. The simulator may generate a simulated traffic scenario including two or more lanes, an ego-vehicle, a dead end position, and one or more traffic participants. The dead end position may be a position by which a lane change for the ego-vehicle may be desired. The simulated traffic scenario may be associated with an occupancy map, a relative velocity map, a relative displacement map, and a relative heading map at each time step within the simulated traffic scenario. The simulator may model the ego-vehicle and one or more of the traffic participants using a kinematic bicycle model. The simulator may build a policy based on the simulated traffic scenario using an actor-critic network. The policy may be implemented on an autonomous vehicle.

Abstract: According to one aspect, systems and techniques for lane selection may include receiving a current state of an ego vehicle and a traffic participant vehicle, and a goal position, projecting the ego vehicle and the traffic participant vehicle onto a graph network, where nodes of the graph network may be indicative of discretized space within an operating environment, determining a current node for the ego vehicle within the graph network, and determining a subsequent node for the ego vehicle based on identifying adjacent nodes which may be adjacent to the current node, calculating travel times associated with each of the adjacent nodes, calculating step costs associated with each of the adjacent nodes, calculating heuristic costs associated with each of the adjacent nodes, and predicting a position of the traffic participant vehicle.

Abstract: According to one aspect, systems and techniques for lane selection may include receiving a current state of an ego vehicle and a traffic participant vehicle, and a goal position, projecting the ego vehicle and the traffic participant vehicle onto a graph network, where nodes of the graph network may be indicative of discretized space within an operating environment, determining a current node for the ego vehicle within the graph network, and determining a subsequent node for the ego vehicle based on identifying adjacent nodes which may be adjacent to the current node, calculating travel times associated with each of the adjacent nodes, calculating step costs associated with each of the adjacent nodes, calculating heuristic costs associated with each of the adjacent nodes, and predicting a position of the traffic participant vehicle.

Abstract: Systems and methods for path planning with latent state inference and spatial-temporal relationships are provided. In one embodiment, a system includes an inference module, a policy module, a graphical representation module, and a planning module. The inference module receives sensor data associated with a plurality of agents. The inference module also maps the sensor data to a latent state distribution to identify latent states of the plurality of agents. The latent states identify agents of the plurality of agents as cooperative or aggressive. The policy module predicts future trajectories of the plurality of agents at a given time based on sensor data and the latent states of the plurality of agents. The graphical representation module generates a graphical representation based on the sensor data and a graphical representation neural network. The planning module generates a motion plan for the ego agent based on the predicted future trajectories and the graphical representation.

Abstract: A system and method for providing probabilistic-based lane-change decision making and motion planning that include receiving data associated with a roadway environment of an ego vehicle. The system and method also include performing gap analysis to determine at least one gap between neighboring vehicles that are traveling within the target lane to filter out an optimal merging entrance for the ego vehicle to merge into the target lane and determining a probability value associated with an intention of a driver of a following neighboring vehicle to yield to allow the ego vehicle to merge into the target lane. The system and method further include controlling the ego vehicle to autonomously continue traveling within the current lane or autonomously merge from current lane to the target lane based on at least one of: if the optimal merging entrance is filtered out and if the probability value indicates an intention of the driver to yield.