Vehicle operation using a dynamic occupancy grid
Methods for operating a vehicle in an environment include receiving light detection and ranging (LiDAR) data from a LiDAR of the vehicle. The LiDAR data represents objects located in the environment. A dynamic occupancy grid (DOG) is generated based on a semantic map. The DOG includes multiple grid cells. Each grid cell represents a portion of the environment. For each grid cell, a probability density function is generated based on the LiDAR data. The probability density function represents a probability that the portion of the environment represented by the grid cell is occupied by an object. A time-to-collision (TTC) of the vehicle and the object less than a threshold time is determined based on the probability density function. Responsive to determining that the TTC is less than the threshold time, a control circuit of the vehicle operates the vehicle to avoid a collision of the vehicle and the object.
1. A method comprising:
receiving, by one or more processors of a vehicle operating in an environment, light detection and ranging (LiDAR) data from one or more LiDARs of the vehicle, the LiDAR data representing one or more objects located in the environment;
generating, by the one or more processors, a dynamic occupancy grid (DOG) by segregating a semantic map of the environment, the DOG comprising a plurality of grid cells, each grid cell of the plurality of grid cells representing a portion of the environment;
for each grid cell of the plurality of grid cells, generating, by the one or more processors, a probability density function based on the LiDAR data in a time-space-velocity coordinate frame, the probability density function representing a probability that the portion of the environment represented by the grid cell is occupied by an object of the one or more objects, wherein the probability density function is a probability density of finding a particle at a respective grid cell at a time t, a location x, and a velocity v, said particle representing the object;
determining, by the one or more processors, that a time-to-collision (TTC) of the vehicle and the object is less than a threshold time based on the time t of the probability density function; and
responsive to determining that the TTC is less than the threshold time, operating, by the one or more processors, the vehicle to avoid a collision of the vehicle and the object.
2. The method of claim 1 , wherein each grid cell of the plurality of grid cells is one of a two-dimensional polygon or a three-dimensional polyhedron.
3. The method of claim 1 , wherein the generating of the dynamic occupancy grid comprises allocating, by the one or more processors, a portion of the LiDAR data to more than one grid cell of the plurality of grid cells, the portion of the LiDAR data corresponding to the object.
4. The method of claim 1 , wherein the generating of the probability density function comprises:
responsive to determining, by the one or more processors, that an intensity of the portion of the LiDAR data corresponding to the object is greater than a threshold intensity, identifying, by the one or more processors, the grid cell representing the portion of the environment.
5. The method of claim 1 , wherein the generating of the probability density function further comprises adjusting, by the one or more processors, the probability that the portion of the environment is occupied by the object to greater than zero.
6. The method of claim 1 , further comprising updating, by the one or more processors, the dynamic occupancy grid using recursive Bayesian analysis on the LiDAR data.
7. The method of claim 1 , wherein the probability density function further represents a first conditional probability that the portion of the environment represented by the grid cell will remain occupied by the object, given that the portion of the environment represented by the grid cell is occupied by the object.
8. The method of claim 1 , wherein the probability density function further represents a second conditional probability that the portion of the environment represented by the grid cell will remain free of the object, given that the portion of the environment represented by the grid cell is free of the object.
9. The method of claim 1 , wherein a length of each edge of the grid cell is in a range from 1 cm to 1 m.
10. The method of claim 1 , wherein the object is a vehicle, the method further comprising generating, by the one or more processors, a second probability density function based on the LiDAR data, the second probability density function representing a second probability that the portion of the environment represented by the grid cell is occupied by a pedestrian, the one or more objects comprising the pedestrian.
11. The method of claim 1 , wherein the LiDAR data comprises noise, the method further comprising:
determining, by the one or more processors, that the probability for a particular grid cell of the plurality of grid cells is greater than zero; and
determining, by the one or more processors, that the probability for neighboring grid cells of the particular grid cell is zero.
12. The method of claim 11 , wherein the LiDAR data comprises noise, the method further comprising adjusting, by the one or more processors, a first probability for the particular grid cell to zero.
13. The method of claim 1 , wherein the generating of the probability density function comprises transforming, by the one or more processors, the LiDAR data into the probability that the portion of the environment represented by the grid cell is occupied by the object using a Fourier transform, the LiDAR data measured in a time domain and the plurality of grid cells generated in a space domain.
14. The method of claim 1 , wherein the generating of the probability density function comprises recursively combining, by the one or more processors, the LiDAR data with a posterior probability that the portion of the environment represented by the grid cell is occupied by the object using a Bayesian filter.
15. The method of claim 1 , wherein the generating of the dynamic occupancy grid comprises segregating the semantic map of the environment into the plurality of grid cells using Cartesian coordinates.
16. The method of claim 1 , wherein the generating of the dynamic occupancy grid comprises segregating the semantic map of the environment into the plurality of grid cells based on polar coordinates.
17. The method of claim 1 , wherein the dynamic occupancy grid is generated based on regular sampling of the semantic map of the environment.
18. The method of claim 1 , wherein the dynamic occupancy grid is generated based on irregular sampling of the semantic map of the environment.
19. A system, comprising:
at least one processor, and
at least one non-transitory storage media storing instructions that, when executed by the at least one processor, cause the at least one processor to:
receive light detection and ranging (LiDAR) data from one or more LiDARs of a vehicle, the LiDAR data representing one or more objects located in an environment;
generate a dynamic occupancy grid (DOG) based on by segregating a semantic map of the environment, the DOG comprising a plurality of grid cells, each grid cell of the plurality of grid cells representing a portion of the environment;
generate a probability density function based on the LiDAR data in a time-space-velocity coordinate frame for each grid cell of the plurality of grid cells, the probability density function representing a probability that the portion of the environment represented by the grid cell is occupied by an object of the one or more objects, wherein the probability density function is a probability density of finding a particle at a respective grid cell at a time t, a location x, and a velocity v, said particle representing the object;
determine that a time-to-collision (TTC) of the vehicle and the object is less than a threshold time based on the time t of the probability density function; and
operate the vehicle in accordance with determining that the TTC is less than the threshold time.
20. The system of claim 19 , wherein each grid cell of the plurality of grid cells is one of a two-dimensional polygon or a three-dimensional polyhedron.