Systems and Controls
Accordions
Requirements generation for simulation-based autonomous vehicles safety assessment
- Developed a way to generate safety-critical driving scenarios that can be simulated using NHTSA CDS crash records data
- Developed an approach to filter, categorize CDS cases
- Used CDS cases to define logical scenarios with a descriptive story of the event
- Developed an optimal control and driver model-based approach to identify scenario parameters such that the scenario becomes marginally critical
Publications/Thesis:
Kibalama, D., Chen, B.S., Tulpule P. ,(2021) AV/ADAS Safety-Critical Testing Scenario Generation from Vehicle Crash Data, SAE World Congress
Chen, B.S., Tulpule P. (2022), Criticality Assessment of Simulation-Based AV/ADAS Test Scenarios, SAE World Congress
Multi-Modal traffic simulation for The Ohio State University campus
- As a part of OSU campus digital twin a procedure to calibrate multi-modal traffic model was developed
- A SUMO and dSpace ASM co-simulation platform was built to simulate traffic in Hardware-in The Loop setting
Thesis:
Kalra, V., (2021) Multi-modal Simulation and Calibration for OSU Campus Mobility, MS Thesis, The Ohio State University
Traffic modeling at roundabouts using infrastructure camera video feed
- An infrastructure camera data fusion method was developed to generate O-D matrix for multi-lane roundabout
- A custom trained YOLO algorithm along with SORT was used to track vehicles across the roundabout
- A Machine Learning-based approach was proposed to generate uncommon behavior at the roundabout
- A calibrated micro traffic simulation model was developed for the roundabout
Publications/Thesis:
Patil, J., Tulpule, P. (2021). Infrastructure Camera Video Data Processing of Traffic at Roundabouts (No. 2021-01-0165). SAE Technical Paper
Patil, J. (2021). Camera Video Data Processing of Traffic and Development of Microsimulation Traffic SUMO Model for Roundabout, MS Thesis, The Ohio State University
Toolset for AV/ADAS validation verification
- A modular open-source software tool was developed for AV/ADAS algorithms V&V
- A SUMO and Unreal (gaming engine) co-simulation platform was developed
- A vehicle dynamics based algorithm allows seamless time synchronization between multi-fidelity simulation/rendering tools
Publications/Thesis:
Tulpule, P., Midlam-Mohler, S., Karumanchi, A., Jin, Y. (2021). A Simulation Tool for Virtual Validation and Verification of Advanced Driver Assistance Systems (No. 2021-01-0865). SAE Technical Paper
Sensitivity analysis of computer vision algorithms
- An Image Quality Assessment (IQA) based novel approach is developed to understand the relationship between image properties and object detection performance
- Various IQAs and performance measures for different detection algorithms were analyzed
- A support vector machine-based classifier was developed to understand relationships between image properties and detection performance
- The approach was verified using various pre-trained models (e.g. YOLO-v3) for the Audi-A2D2 dataset
Publications/Thesis:
Padisala, S.K., (2021) Development of Frameworks for Environment Dependent Traffic Simulation and ADAS Algorithm Testing, MS Thesis, The Ohio State University