RETURN TO PORTFOLIO
Industry InternshipJul 2025 – Aug 2025Chennai, India

STEMTEC AI & Robotics Pvt. Ltd.

AI & Robotics Intern — Indoor Autonomous Mobile Robot

Worked on the software stack of an indoor autonomous mobile robot, mostly on the sensor validation and data analysis side rather than pure hardware work. I ran time series error analysis on Extended Kalman Filter telemetry at 15 Hz against raw encoder logs across repeated ground truth trials, which brought mean spatial error down 60 percent and angular error down about 69 percent. Also worked with high frequency LiDAR data, close to 3,600 points a second, to validate a filtering approach for obstacle detection, getting it to 95 percent accuracy across 20 stress test trials while keeping latency under 50ms. A good chunk of the work was really about designing solid test trials and making sure the numbers I was reporting actually held up under scrutiny.

SPATIAL DRIFT REDUCTION
60.0%

Mean spatial drift cut from 0.45 m → 0.18 m across 10 m ground-truth trials via 15 Hz EKF.

ANGULAR ERROR MITIGATION
69.2%

Rotational yaw orientation error reduced from 7.8° → 2.4° fusing IMU + encoder odometry.

NAVIGATION SUCCESS
95.0%

Collision-free obstacle traversal across 20 randomized physical stress tests (19/20 successful passes).

TELEMETRY THROUGHPUT
<50 ms

Sub-50 ms end-to-end control latency ingesting ~3,600 LiDAR points/sec at 10 Hz frequency.

Hardware & Simulation Video Feeds

Live recordings from physical chassis validation trials and ROS 2 simulation suites.

ROS 2 Jazzy — SLAM & Autonomous Navigation
Open in Drive
RViz 2 scan visualization showcasing real-time costmap inflation, 2D LiDAR /scan filtering, and dynamic trajectory computation.View Full HD on Drive
Physical Chassis — Working Demonstration
Lab Corridor Trial
Differential-drive physical bot traversing lab corridors with active LiDAR safety cutoff engaged and 15 Hz Extended Kalman Filter state telemetry.

ROS 2 Software Architecture

Modular publisher/subscriber nodes implemented in Python 3 under the ROS 2 Jazzy framework.

robot_localization (EKF)

State-Space Estimation

Fuses asynchronous 50 Hz wheel encoder odometry (/odom) with 30 Hz 6-DOF IMU acceleration and gyro telemetry (/imu/data). The continuous Extended Kalman Filter suppresses cumulative wheel slip drift by 60% on low-friction flooring.

Frequency: 15 Hz • TF: odom → base_link
obstacle_avoider Node

Frontal Sector Slicing

Subscribes to 360° 2D LiDAR range data on /scan. Employs circular modulo array-indexing to isolate a 60° forward cone (±30° from bore). If range minimum falls below the 0.75 m safety threshold, an emergency angular override is injected into the velocity multiplexer.

Latency: <50 ms • Range Threshold: 0.75 m
two_point_looper Node

Patrol Trajectory Loop

Maintains timed autonomous patrol cycles between predefined coordinate waypoints. Publishes linear and angular velocity commands to /cmd_vel at 0.4 m/s nominal traverse velocity with smooth ramp deceleration curves on waypoint approach.

Traverse Speed: 0.4 m/s • Timer Rate: 10 Hz

Physical Hardware vs Gazebo Benchmark

Telemetry comparison between digital twin simulation runs and physical laboratory hardware trials.

Evaluation MetricRaw Baseline (No Filter)Gazebo SimulationPhysical Robot ChassisNet Improvement
Mean Spatial Drift (10m)0.45 m0.08 m0.18 m-60.0% drift
Angular Heading Error7.8°1.1°2.4°-69.2% error
Obstacle Avoidance Rate65% (13/20)100% (20/20)95% (19/20)+30.0% success
End-to-End Latency110 ms22 ms46 msSub-50 ms target

Telemetry & Simulation Evidence

Click any screen capture below to inspect full-resolution ROS 2 graphs, costmaps, and terminal telemetry.

Local dashboard showing live camera feed, obstacle/person detection and face detection results
Dashboard + CV Pipeline
Local dashboard showing live camera feed, obstacle/person detection and face detection results
ROS 2 Gazebo simulation showing the rover performing autonomous patrol movement in a warehouse environment
ROS 2 Gazebo Patrol Simulation
ROS 2 Gazebo simulation showing the rover performing autonomous patrol movement in a warehouse environment
RViz visualization of initial rover motion with 2D LiDAR point-cloud based navigation
RViz LiDAR — Initial Motion
RViz visualization of initial rover motion with 2D LiDAR point-cloud based navigation
RViz visualization showing LiDAR-based obstacle avoidance and turning during patrol
RViz LiDAR — Turning Motion
RViz visualization showing LiDAR-based obstacle avoidance and turning during patrol