Siddhant Shete

Engineered a real-time anomaly detection framework using TFLite and ONNX, successfully deploying it on mobile robots with ROS and Nvidia Jetson for seamless integration. Developed RNN models for sequence-to-sequence forecasting with IMU data, enabling tip-over prevention in exploration vehicles via the ROCK framework. Currently designing an obstacle avoidance algorithm for lunar micro-rovers, targeting FPGA deployment for efficiency. Skilled in team management and supervising research assistants effectively

• Developed deep learning algorithms for condition monitoring for fluid power systems to improve system performance and reliability. • Researched and evaluated innovative machine learning technologies, explaining details to how the feature selection takes place for automatic parameter adjustments.

• Designed and developed deep learning vision-based obstacle detection algorithm with a focus on anomaly detection in autonomous vehicles. • Optimized the algorithm for context-dependent anomaly detection with regard to its robustness. • Conducted research with regard to suitability for safety-critical applications.

• Developed a model for fish detection and classification according to size and shape using YOLO v4. • Tracking of fishes using optical flow and categorized them according to type.

• Scanning and extraction of the 3D point cloud data for robotic operations and collision avoidance. • Robot programming and Simulation for automotive project in the area of Digital Twin.