An innovative control suite, compatible with streaming and gymnasium environments, that standardizes the management of vehicles and sensor reading for any agents, both in simulated and real-world situations. The project is presently undergoing active development, and is utilized by the ROAR for both research and instructional purposes.

• GitHub

A basic, yet comprehensive, database implementation designed to effectively execute transactions. Advanced features such as B+ tree indices, effective join algorithms, and sophisticated query optimization have been successfully deployed and integrated. Moreover, multigranularity locks are introduced to enable concurrent execution of transactions, enhancing the overall system performance. Importantly, the implementation also includes database recovery mechanisms, ensuring data security and reliability in the face of potential system failures. With these features, RookieDB offers a robust and comprehensive platform for database management and operation.

• GitHub

A comprehensive project that seamlessly stitches images together through a blend of manual and automatic feature selection methods, alongside various blending techniques. Starting with manually selected feature points for homography transformations, it evolves to adopt automatic feature selection algorithms, using Harris Corners Detection to identify feature points and Adaptive Non-Maximal Suppression to distill the most crucial ones. Each feature point is then described by a Feature Descriptor that encompasses its neighboring pixel information for matching. The matching procedure is facilitated by Random Sample Consensus (RANSAC) to determine the most fitting set of matching features, subsequently utilized to calculate the final homography matrix for image warping and blending.

• GitHub

This cloth simulator project is a real-time simulation built using a mass-spring system. Initialization includes constructing a grid of masses and springs to represent cloth, then computing forces on each point mass and updating their positions using Verlet integration. Collisions with other objects like spheres and planes are handled, and self-collision techniques are implemented using a hash table to prevent unrealistic cloth behavior. The cherry on top includes the introduction of various shading techniques for enhanced visual aesthetics, including diffuse and Blinn-Phong shading, texture mapping, displacement and bump mapping, and environment-mapped reflections. The result is a comprehensive, realistic cloth simulator suitable for diverse applications.

• GitHub