Gopisainath Mamindlapalli

About Gopisainath Mamindlapalli

I am a robotics engineer building and deploying a quadruped robot from scratch at Rainier Labs. I own the full system end to end: mechanical design, electronics, firmware and ROS 2-based software running on embedded hardware. My technical depth is in.I build simulation environments (MuJoCo, IsaacLab), train locomotion policies, and transfer them to physical hardware while handling real-world constraints such as actuator limits, latency, noise, and model mismatch. This work runs on an actual robot, not just in simulation.I build and deploy localization stacks (VIO/LIO), calibrate and time-sync sensors, fuse multi-sensor odometry using EKF-based sensor fusion, and ship SLAM/mapping pipelines that support reliable navigation. I’ve also developed 3D reconstruction pipelines using voxel-grid occupancy maps and Structure-from-Motion (SfM).I work across Python and C++ and routinely bridge low-level systems (motor drivers, firmware, CAN, controllers) with higher-level autonomy and learning pipelines. My mechanical background lets me reason about dynamics, actuation, and failure modes, not just software abstractionsI am for teams building, especially legged systems, robot learning for deployment, and full-stack robotics engineering.

Experience

1. Lead Robotics Engineer (Founding Engineer)

   Rainier Labs

   Jul 2024 — Present · San Jose, CA, US

   Deployed VIO on a real robot with RealSense D435i+IMU, publishing odom at 30Hz for Nav2 and maintaining continuous localization across multiple indoor loops.• Fused proprioceptive base odometry (IK/WBC) with visual odometry via an EKF (robot localization pkg), producing a stable odom to base link state estimate that improved navigation robustness.• Productionized indoor Nav2 by generating 2D LiDAR maps with Cartographer and localizing with AMCL, feeding the EKF-fused odometry as the odometry input to achieve 10sec pose convergence after initialization.• Built a bag-replay evaluation pipeline to measure closed-loop drift; achieved ±10 cm end-to-start error over half mile loops and reduced VIO dropouts by fixing motion blur and feature loss with exposure tuning and motion gating.• Calibrated camera intrinsics/extrinsics using OpenCV, characterized IMU noise and bias with Allan variance, estimated camera–IMU extrinsics and time offset with kalibr + online refinement, reducing tracking failures during fast motion.• Implemented 2D LiDAR ICP odometry and reduced drift using loop-closure + GTSAM pose-graph optimization; improved long-run map consistency and trajectory closure vs. raw ICP.• Tuned Nav2 costmaps and DWB local planner by adjusting footprint, inflation and obstacle layers, velocity and acceleration limits to eliminate spin-in-place failures at doorways, achieving reliable autonomous runs at up to 1 m/s.• Hardened autonomy stack for demos: containerized builds, added launch/regression tests, runtime health/diagnostics, and recovery behaviors; improved field reliability and reduced operator intervention across 25+ public deployments.• Owned end-to-end development of the quadruped including mechanical design(OnShape, Solidworks), fabrication(FDM, SLA, SLS and CNC Machining), electrical assembly, firmware coding (UART, CAN protocols), motor calibration.

Education

• Osmania University, Hyderabad

  Bachelor's degree, Mechanical Engineering

• Northeastern University

  Master of Science - MS, Mechatronics, Robotics, and Automation Engineering