Autonomously guiding vehicles using vehicles-centered occupancy grid including collected information about road users
A method for providing information detected by on-board sensors about road users in the surroundings of a vehicle involves providing the detected information as data structures representing vectors. Every vector is thereby associated with a cell of a predefined vehicle-mounted occupancy grid, in which a road user described by the respective information is situated. Every vector at least includes coordinates of an associated cell, in which the respective road user is situated, a speed vector, which represents a speed of the respective road user, a time stamp, which represents a point in time of a detection of the respective road user, and an object class, which represents a type of the respective road user.
1 . A method, comprising:
detecting, by on-board sensors of a plurality of vehicles, information about a road users in surroundings of the plurality of vehicle;
generating, by each of the plurality of vehicles based on respective detected information, a vehicle-centered occupancy grid comprising a set of cells centered on a respective one of the plurality of vehicles, wherein cell each cell of the vehicle-centered occupancy grid is occupied or is unoccupied, wherein occupied cells include a vector for one of the road users detected in the respective occupied cell, wherein the vector comprises
position of the one of the detected road users relative to the respective one of the plurality of vehicles;
a speed vector of a speed of the one of the road users detected in the respective occupied cell,
a time stamp of when a respective on-board sensor of a respective one of the plurality of vehicles detected the one of the road users detected in the respective occupied cell,
an orientation of the one of the road users detected in the respective occupied cell, and
an object class of a type of the one of the road users detected in the respective occupied cell;
transmitting, by the plurality of vehicle, a respective vehicle-centered occupancy grid to a back-end server;
converting, by the back-end server, each of the respective vehicle-centered occupancy grids into a global occupancy grid defined by a global coordinate system;
transmitting, by the back-end server to an autonomous vehicle that is one of the plurality of vehicles, the global occupancy grid;
converting, by the autonomous vehicle, the global occupancy grid into an occupancy grid centered on the autonomous vehicle, wherein the occupancy grid centered on the autonomous vehicle includes
cell beyond a range of at least one sensor of the autonomous vehicle; or
cells that are not in a field of vision of the autonomous vehicle;
generating, by the autonomous vehicle based on the occupancy grid centered on the autonomous vehicle, a safe driving path; and
autonomously guiding the autonomous vehicle along the safe driving path.
2 . The method of claim 1 , further comprising:
generating, by the autonomous vehicle, a route; and
autonomously guiding the autonomous vehicle along the route.
3 . The method of claim 1 , further comprising:
comparing, by the autonomous vehicle, a sensor reading corresponding to a road user detected by the autonomous vehicle with a vector at a corresponding cell in the occupancy grid centered on the autonomous vehicle; and
determining, by the autonomous vehicle based on the comparison, that a sensor that generated the sensor reading is functionally impaired.
4 . The method of claim 1 , further comprising:
predicting, by the autonomous vehicle using the vector, behavior of the one of the detected road users.
5 . The method of claim 1 , wherein position of the one of the detected road users relative to the respective one of the plurality of vehicles is a three-dimensional position.
6 . The method of claim 1 , wherein road users in surroundings of the plurality of vehicles include other vehicles and pedestrians.
7 . The method of claim 1 , wherein on-board sensors of the plurality of vehicles include at least one radar, lidar, camera, and ultrasound sensor.
8 . The method of claim 1 , wherein the vector further comprises a detection reliability of the detection of the one of the detected road users.
9 . The method of claim 1 , wherein unoccupied cells are set to a value of zero.
10 . The method of claim 1 , wherein each of the plurality of vehicles tracks the detected road users over time and generate new vectors for the detected road users over time.
11 . The method of claim 1 , wherein the global occupancy grid is a city-wide occupancy grid.
12 . The method of claim 1 , wherein the occupancy grid centered on the autonomous vehicle is enhanced with safety-critical information.
13 . The method of claim 12 , wherein the safety-critical information includes road signs recognized by one of the plurality of vehicles or unusual stationary objects, wherein the unusual stationary objects are one of a double-parked vehicle, a crashed vehicle, and lost cargo.
14 . The method of claim 12 , wherein the occupancy grid centered on the autonomous vehicle is enhanced with non-critical information.
15 . The method claim 14 , wherein the non-critical information includes one of information about occupied parking spaces and potholes.