Systems and methods for situational behavior of an autonomous vehicle are disclosed. In one aspect, an autonomous vehicle includes at least one perception sensor configured to generate perception data indicative of at least one other vehicle on a roadway, a non-transitory computer readable medium, and a processor. The processor is configured to determine that the other vehicle is violating one or more rules of the roadway based on the perception data, tag the other vehicle as a non-compliant driver, and modify control of the autonomous vehicle in response to tagging the other vehicle as a non-compliant driver.

Disclosed are distributed computing systems and methods for controlling multiple autonomous control modules and subsystems in an autonomous vehicle. In some aspects of the disclosed technology, a computing architecture for an autonomous vehicle includes distributing the complexity of autonomous vehicle operation, thereby avoiding the use of a single high-performance computing system and enabling off-the-shelf components to be use more readily and reducing system failure rates.

Disclosed herein is a technique for generating and providing an indication to an autonomous vehicle regarding the confidence level for the accuracy or quality of the map data in which the indication is determined from observation data received from other vehicles.

Embodiments of this application provide a map data collection method and apparatus, and a system, to report map data in a targeted manner. The method includes: receiving a first instruction from a network side device, where the first instruction instructs a map data reporting manner to a first vehicle, the first instruction includes confidence information, and the confidence information indicates confidence that map data reported by the first vehicle; and sending the map data to the network side device in the map data reporting manner instructed by the first instruction, where confidence of the map data is not lower than the confidence indicated by the confidence information.

A method for constructing a lane-level navigation map is provided. The method may include: determining a change position of a lane in a road-level electronic map; extracting lane information of the change position of the lane, the lane information including a lane connectivity relationship; and storing the lane information in the road-level electronic map to obtain the lane-level navigation map.

Provided are systems and methods for controlling a vehicle based on a map that designed using a factor graph. Because the map is designed using a factor graph, positions of the road can be modified in real-time while operating the vehicle. In one example, the method may include storing a map which is associated with a factor graph of variable nodes representing a plurality of constraints that define positions of lane lines in a road and factor nodes between the variable nodes on the factor graph which define positioning constraints amongst the variable nodes, receiving an indication from the road using a sensor of a vehicle, updating positions of the variable nodes based on the indication and an estimated location of the vehicle within the map, and issue commands capable of controlling a steering operation of the vehicle based on the updated positions of the factor nodes.

Generating a map associated with an environment may include collecting sensor data received from one or more vehicles and generating a set of links to align the sensor data. A mesh representation of the environment may be generated from the aligned sensor data. A system may determine a proposed link to add, a proposed link deletion, and/or a proposed link alteration, and receive a modification comprising instructions to add, delete, or modify a link. Responsive to receiving a modification, the system may re-align a window of sensor data associated with the modification. The modification and/or sensor data associated therewith may be collected as training data for a machine learning model, which may be trained to generate link modification proposals and/or determine sensor data that may be associated with a poor sensor data alignment.

A method of generating a roadway map includes receiving an image of a roadway. The method further includes performing a spectral analysis of the received image to determine reflectivity data for a plurality of wavelengths of light. The method further includes identifying a feature of the roadway in response to the determined reflectivity data exhibiting a reflection peak. The method further includes classifying the identified feature based on a size or a pitch of the exhibited reflection peak. The method further includes generating the roadway map based on the classification of the identified feature.

A method of generating a first person view map includes receiving an image from above a roadway. The method further includes generating a road graph based on the received image, wherein the road graph comprises a plurality of road segments. The method further includes converting the received image using the road graph in order to generate a first person view image for each road segment of the plurality of road segments. The method further includes combining the plurality of road segments to define the first person view map.

An end-to-end system for data generation, map creation using the generated data, and localization to the created map is disclosed. Mapstreams—or streams of sensor data, perception outputs from deep neural networks (DNNs), and/or relative trajectory data—corresponding to any number of drives by any number of vehicles may be generated and uploaded to the cloud. The mapstreams may be used to generate map data—and ultimately a fused high definition (HD) map—that represents data generated over a plurality of drives. When localizing to the fused HD map, individual localization results may be generated based on comparisons of real-time data from a sensor modality to map data corresponding to the same sensor modality. This process may be repeated for any number of sensor modalities and the results may be fused together to determine a final fused localization result.