ViaBot Inc.
Robotics Deployment Intern
Designing and optimizing navigation maps for autonomous robots
Project Overview
During my internship at ViaBot, I focused on working with navigation maps for autonomous robots deployed at high-profile client locations including Google HQ and Adobe. My work involved learning how to interpret complex 3D spatial data and understanding how it's transformed into 2D navigation maps that enable robots to safely navigate through various environments.
I gained hands-on experience with LiDAR, depth cameras, and various mapping technologies while assisting with navigation solutions that handle the complexities and edge cases of real-world environments. This internship provided valuable exposure to both the technical aspects of robotics and practical problem-solving in the field.
My Role
- •Map Testing & Analysis
- •Sensor Data Interpretation
- •Navigation Testing
- •ROS Exploration
- •Client Site Support
Technologies Used
The Challenge
Autonomous robot navigation in complex, dynamic environments presents several significant challenges:
- 1Complex Environments: Client locations like Google HQ feature diverse spaces with varying obstacles, elevators, and high-traffic areas
- 2Sensor Limitations: Each sensor type (LiDAR, depth cameras, RGB cameras) has inherent limitations and blind spots
- 3Dynamic Obstacles: People, furniture, and other moving objects constantly change the environment
- 4Edge Cases: GPS signal loss, reflective surfaces, and lighting variations create navigation challenges
Mapping Process
3D Data Collection
During my internship, I learned about the first step in creating effective navigation maps: collecting comprehensive 3D spatial data of client environments. This process involved:
- •Learning how LiDAR sensors are configured and calibrated for optimal data collection
- •Understanding how depth cameras capture detailed spatial information
- •Assisting with systematic scans of client locations to ensure complete coverage
- •Observing how point cloud data is processed and cleaned to remove noise and artifacts
2D Map Conversion
After collecting 3D data, I gained hands-on experience with the techniques used to convert this information into optimized 2D navigation maps that robots could efficiently use:
- •Working with flattened 3D point clouds while learning how critical navigation information is preserved
- •Studying cost maps that represent areas of varying navigation difficulty
- •Helping identify permanent obstacles, restricted zones, and preferred paths
- •Learning about map resolution optimization to balance detail with computational efficiency
Map Optimization & Testing
The final phase involved testing the maps in real-world conditions, where I gained valuable experience:
- •Using RViz to visualize navigation parameters and understand their effects
- •Learning how C++ scripts handle edge cases and improve navigation algorithms
- •Participating in on-site testing with robots in client environments
- •Observing how maps are iteratively refined based on real-world performance data
Client Deployments
Google Headquarters
Mountain View, CA
During my internship, I had the opportunity to assist with the Google HQ deployment, which presented unique challenges due to its large campus with multiple buildings, open spaces, and high foot traffic. I learned about:
- •How interconnected maps enable seamless navigation between buildings
- •Techniques for optimizing robot movement in high-traffic areas
- •Special handling methods for glass walls and reflective surfaces
Adobe Campus
San Jose, CA
I also supported the Adobe deployment, which involved a multi-floor office environment with complex interior layouts. During this experience, I gained insights into:
- •How floor-specific maps with elevator transition points are utilized
- •Navigation parameters that help robots navigate narrow corridors
- •Time-based navigation rules implemented for different office hours
Ready to see more?
Check out my other projects or get in touch.