Fleet utilization efficiency for on-demand transportation services
An on-demand transportation management system can collect vehicle fleet utilization data corresponding to human-driven vehicles (HDVs) and autonomous vehicles (AVs) operating within a given region in connection with an on-demand transportation service. The on-demand transportation management system can then establish a set of selection priorities for respective areas of the given region based on the vehicle fleet utilization data, each selection priority indicating whether a respective area of the given region is to favor HDVs or AVs for servicing transport requests.
1. A computing system comprising:
one or more processors; and
one or more memory resources storing instructions that are executable by the one or more processors to cause the computing system to:
receive a new unverified software version that relates to operation of autonomous vehicles;
generate a simulation for pre-certifying the new unverified software version, wherein the simulation is generated using a previous verified software version and real-world log data from at least one of the autonomous vehicles,
wherein the real-world log data comprises at least one of: (i) recorded sensor data, (ii) telemetry data, or (iii) diagnostic data obtained while the at least one autonomous vehicle implemented the previous verified software version;
execute one or more implementations of the simulation for the new unverified software version; and
pre-certify the new unverified software version for execution on the autonomous vehicles by verifying outputs of the new unverified software version against a comparative simulation of a human driver.
2. The computing system of claim 1 , wherein the new unverified software version relates to operation of the autonomous vehicles within a given geographic region.
3. The computing system of claim 1 , wherein the simulation is also generated using real-world log data collected through a human driven vehicle.
4. The computing system of claim 1 , wherein the instructions are executable by the one or more processors to further cause the computing system to:
adjust a simulation parameter of the simulation; and
execute one or more additional implementations of the simulation.
5. The computing system of claim 4 , wherein the simulation parameter is adjusted to include an additional actor.
6. The computing system of claim 5 , wherein the additional actor comprises at least one of: (i) a vehicle, or (ii) a pedestrian.
7. The computing system of claim 1 , wherein executing the simulation further comprises generating a simulation result.
8. The computing system of claim 7 , wherein the instructions are executable by the one or more processors to further cause the computing system to:
verify, based on the simulation result, the new unverified software version for at least one of: (i) further simulation, or (ii) real-world use.
9. The computing system of claim 1 , wherein the new unverified software version is executable by computational resources of an autonomous vehicle to simulate autonomously operating control mechanisms of the autonomous vehicle based on a simulated sensor view.
10. The computing system of claim 1 , wherein the instructions are executable by the one or more processors to further cause the computing system to:
compare data of the one or more implementations of the simulation for the new unverified software version to data of one or more implementations of a simulation of a previously verified software version.
11. The computing system of claim 10 , wherein to pre-certify the new unverified software version for execution on the autonomous vehicles the computing system is further caused to:
determine that the data of the one or more implementations of the simulation for the new unverified software version and the data of the one or more implementations of the simulation of the previously verified software version are within a nominal range indicating that the new unverified software version performs as well as or better than the previous verified software version for its stated purposes, and in response, pre-certify the new unverified software version for distribution to the autonomous vehicles.
12. The computing system of claim 1 , wherein to pre-certify the new unverified software version the computing system is caused to:
compare data of the one or more implementations of the simulation for the new unverified software version to data of one or more implementations of a simulation of a human driven vehicle.
13. The computing system of claim 12 , wherein to pre-certify the new unverified software version for execution on the autonomous vehicles the computing system is further caused to:
determine that the data of the one or more implementations of the simulation for the new unverified software version and the data of the one or more implementations of the simulation of the human driven vehicle are within a nominal range indicating that the new unverified software version performs as well as or better than the previous verified software version for its stated purposes, and in response, pre-certify the new unverified software version for distribution to the autonomous vehicles.
14. The computing system of claim 1 , wherein the autonomous vehicles comprise fully autonomous self-driving vehicles.
15. A method comprising:
receiving a new unverified software version that relates to operation of autonomous vehicles;
generating a simulation for pre-certifying the new unverified software version, wherein the simulation is generated using a previous verified software version and real-world log data from at least one of the autonomous vehicles,
wherein the real-world log data comprises at least one of: (i) recorded sensor data, (ii) telemetry data, or (iii) diagnostic data obtained while the at least one autonomous vehicle implemented the previous verified software version;
executing one or more implementations of the simulation for the new unverified software version; and
pre-certifying the new unverified software version for execution on the autonomous vehicles based on a result of the one or more implementations meeting a predetermined standard by verifying outputs of the new unverified software version against a comparative simulation of a human driver.
16. The method of claim 15 , wherein executing one or more implementations of the simulation for the new unverified software version comprises:
executing a plurality of implementations of the simulation for the new unverified software version, wherein after a first implementation of the simulation the new unverified software version does not perform in accordance with the predetermined standard, generating a targeted simulation report;
debugging the new unverified software version;
executing a second implementation of the simulation; and
determining a performance of the new unverified software version as compared to one or more previously verified software versions.
17. The method of claim 15 , further comprising:
verifying the new unverified software version for at least one of: (i) further simulation, or (ii) real-world use.
18. The method of claim 15 , further comprising:
adjust a simulation parameter of the simulation, wherein the simulation parameter comprises an object within the simulation; and
execute one or more additional implementations of the simulation.
19. One or more non-transitory computer readable media that store instructions that are executable by one or more processors to perform operations comprising:
receiving a new unverified software version that relates to operation of autonomous vehicles;
generating a simulation for pre-certifying the new unverified software version, wherein the simulation is generated using a previous verified software version and real-world log data from at least one of the autonomous vehicles,
wherein the real-world log data comprises at least one of: (i) recorded sensor data, (ii) telemetry data, or (iii) diagnostic data obtained while the at least one autonomous vehicle implemented the previous verified software version;
executing one or more implementations of the simulation for the new unverified software version; and
pre-certifying the new unverified software version for execution on the autonomous vehicles based on a result of the one or more implementations meeting a predetermined standard by verifying outputs of the new unverified software version against a comparative simulation of a human driver.
20. The one or more non-transitory computer readable media of claim 19 , wherein pre-certifying the new unverified software version comprises:
comparing data of the one or more implementations of the simulation for the new unverified software version to data of one or more implementations of a simulation of a human driven vehicle, and in response, pre-certifying the new unverified software version.