top of page

I'm

chezhian

Visvajiit

Scroll down to find more about me

About Me.

I'm a robotics researcher passionate about autonomous mobile robots. I envision a future where my drone picks up my groceries at the push of a button. The idea of autonomy in vehicles and robots excites me, and I want to contribute to this field. My areas of specialization are Trajectory Planning and SLAM.

​

Outside of research, I'm a gym rat and a trained bachata dancer.

Publications

V. Chezhian and S. Rao, "A Vectorial Approach to Particle Filter Weighting and Resampling for Robot Localization," 2025 25th International Conference on Control, Automation and Systems (ICCAS), Incheon, Korea, Republic of, 2025, pp. 1457-1462, DOI: 10.23919/ICCAS66577.2025.11301265

Work Experience

OCT 2022 - JULY 2025

Research Associate
Ascend Studios
International Institute of Information Tecnology Bangalore
Bengaluru

Education

2018 - 2022

CGPA : 8.48
Bachelor of Technology - Mechanical Engineering
Manipal Institute of Technology
Manipal

2014 - 2018

12th : 90.4 % (CBSE)
10th: GP= 10/10 (CBSE)
Suguna PIP School
Coimbatore

PROJECTS

Zero-clamped C2-continuous Jerk-Minimized Quintic Time Trajectories
  • Implemented a path smoothing algorithm using quintic time trajectories to achieve C2 continuous global trajectories for a self-navigating ground robot.  
  • Developed quintic blending for seamless transitions between linear path segments generated by Orientation Aware A* algorithm. 
  • Formulated an analytical solution that simultaneously bounds blend times and enforces convex‐hull‐constrained trajectories, replacing traditional iterative feasibility and collision‐detection procedures.
  • Investigating optimal start/end point selection for jerk‐minimized blending to improve trajectory smoothness and dynamic performance.
Vectorial approach to Particle Filter weighting
•Developed a novel vector-based weighting method for particle filters, improving robustness to outliers and symmetric map ambiguities in global localization.  
•Achieved more successful global localization by replacing the standard sensor model implementation with proposed method in localization pipelines.
• Authored a first‐author research paper on this approach, accepted for publication at ICCAS 2025
Cross-Sensor Point Cloud Registration
​
•Implemented cross-sensor point cloud matching using ArUco markers, integrating camera and depth sensor data 
•Developed a robust method for detecting and estimating the pose of ArUco markers, enhancing the precision of 3D mapping.
•Achieved seamless point cloud alignment, improving the efficiency and accuracy of multi-sensor fusion in complex environments.
Multi-agent path planning using HCA*

•Implemented Hierarchical Cooperative A* (HCA*) for a warehouse management-oriented startup.
•Generated collision-free multi-agent trajectories, optimizing the efficiency and safety of warehouse operations.
Orientation aware modified A* algorithm
•Developed an orientation-aware A* algorithm respecting non-holonomic constraints of ground robots.
•Optimized the search space by biasing easier motions, narrowing down to more likely waypoints and reducing the number of steps to reach the goal.
•Implemented a method to prevent sharp turns, aiding higher curvature blends.
•Achieved faster cornering as a result of the implemented method.
Localization and Autonomous Navigation
•Achieved real-time global localization using particle filter with mean error of 10 cm and 0.4 rad in a 13x15 meter room on all successful attempts using MATLAB.
•Implemented the A* algorithm for path planning
•Applied piecewise-stitched cubic time trajectories using obtained waypoints, ensuring global velocity continuity and piecewise acceleration continuity.
Rhythmic flight with Quadcopter
•Achieved music-synchronised quadcopter movement in a confined three-dimensional space using MATLAB.
•Utilized Fast Fourier Transform to identify prominent frequencies and track their intensity changes over the piece of music.
•Identified patterns in varying intensities over multiple octaves to determine the time to traverse segments.
•Applied 1-dimensional cubic time trajectories to ensure smooth velocity profiles.

Miscellaneous 

Cubic Bezier Trajectory Tracking

Three drone hover

Outreach

Click icon for my CV

bottom of page