Projects

Project F 1/10th - UIUC

Perception: Implemented precise lane following for the vehicle to stay within its designated lane with a cross track error of less than 5%.
Implemented a pure pursuit controller for precise lateral control, coupled with a PID system for effective longitudinal control.
Developed and integrated an obstacle avoidance system to detect and respond to unexpected obstacles in real-time.

Developed autonomous exploration robot with object detection, navigation and exploration algorithms for search and rescue operations.
Utilized CMU's TARE planner for exploration, integrated YOLOv8 for object detection, employed depth images-based goal estimation & navigation.
Successfully tested in various unfamiliar environments including garage, forest, and indoor, achieving 100% goal detection in all.
During goal (humans, cars, etc.) search and rescue tests, the robot successfully reached targets within a 2-meter radius in all three environments.

Developed an automatic gearbox for 3-wheeler EV, optimizing parameters with genetic algorithms for maximum efficiency and minimum weight.
Implemented helical gears, designing the gearbox to resist bending and pitting failures in gear teeth and achieved an efficiency of 96%.
Introduced spoked gearbox design which further reduced weight from 15.79 kg to 10.21 kg (by 36%) from the initial solid design.
Achieved smooth vehicle performance at high efficiency across diverse road profiles and drive cycles.
Verified theoretical stress values from genetic algorithm design using ANSYS FEA simulations.

Developed two 9 DOF math models of 3-wheeler using Mathematica for vehicle dynamics analysis, utilizing Lagrangian equations of motion.
Utilized robust DOE & Taguchi optimization methods to estimate suspension parameters under random road excitations
Optimized suspension parameters and achieved Sperling ride comfort and quality indices of 1.5 (rated very good) across diverse road profiles and drive cycles.
Harvested 60% of energy previously lost as heat using the regenerative suspension system on par with highly efficient regenerative systems available in the market.

A parametric study of a 3-wheeler passive suspension system vehicle dynamics model created in Mathematica which evaluates ride indices
Obtained a 99% match of simulation results of the Mathematica model and a COMSOL model thus confirming the mathematical model accuracy
The final ride comfort & ride quality indices solved for various ride cycles and terrains were obtained as "GOOD" as per ISO ride evaluation scales