Deliverable #1 - Develop specified simulation test scenarios and scripts for a specified route.
The major objective of this task is to prepare each team to get familiarized with LGSVL simulator and Baidu Apollo platform. Each team must select maps and vehicles provided by the LGSVL simulator, and generate six Python scenario scripts for selected six test sample scenarios listed in the following six categories. A simulation execution report (for scenario scripts) must be generated using the LGSVL simulator and submitted to the specified group fold in the given submission link.
- Scenarios and scripts
- Simulation execution report (generated using the LGSVL simulator)
- Scenario demo videos
- Selected scenario trip/route and map information
The six sample test scenarios will be selected from the following six categories.
1. Perform lane change/low-speed merge
2. Perform vehicle following
3. Move out of travel lane/park
4. Detect and respond to school buses
5. Detect and respond to encroaching oncoming vehicles
6. Detect and respond to pedestrians
Reference: “A Framework for Automated Driving System Testable Cases and Scenarios” The United States National Highway Traffic Safety Administration.
Similar to AV road testing, AV simulation testing is a very complex task due to the following challenges:
- Lack of cost-effective automation tool for test scenario and script generation
- Lack of well-defined adequate validation standards and criteria
The challenge task in the second phase is to ask each team to do their best to select the map(s) and driving route(s) to use effective approaches to generate diverse AV simulation test scenarios and scripts. There are two major objectives: a) achieve good scenario diversity, and b) detect AV problems in the simulation environment.
During this phase, each team is going to create and submit diverse simulation test scenarios and scripts using effective approaches based on the criteria listed in the “Evaluation Criteria” section. Teams can create scenarios based on the reports published by National Highway Traffic Safety Administration.
· Scenarios and scripts
· One simulation execution report for all scenario scripts. This execution report must be generated using LGSVL simulator.br
· Scenario demo videos
· Selected scenario trip/route and map information
· A simulation test report, which reports selected routes with maps, generated test scenarios in classes, scripting methods, simulation test coverage, and problem findings.
 Pre-Crash Scenario Typology for Crash Avoidance Research
 A Framework for Automated Driving System Testable Cases and Scenarios
Registration deadline :
March 15 2021
Training, preparation starts:
First phase challenge submission:
April 30 2021
Notification of selected teams :
May 15 2021
Second phase challenge submission:
July 15 2021
July 15 2021
AI Test Conference presentations
August 23 2021
Announcement of winners: