A Curious Journey into Robotics

Swapnil Meshram

Hey there! I am an Electronics and Robotics Engineer with an interdisciplinary background spanning hardware, software, robotics, and machine learning. I really enjoy working at the intersection of these areas, particularly in understanding how autonomous systems can perceive and interact with the world around them.

This curiosity started when I was 14. I was fascinated by how infants learn just by observing the world around them, which left me wondering if machines could develop a similar curiosity. During my undergraduate years, I began exploring these ideas more seriously by designing robot circuitry, writing embedded firmware, and building robots from scratch. The more I built, the more I wanted to understand how a robot could go beyond following instructions and actually make sense of what was around it.

Today, after completing my master's degree in Robotics and Autonomous Systems at Arizona State University, my early curiosity has evolved into a focused research interest in autonomous systems and machine learning. My research interests are centered around leveraging deep neural networks to learn visual representations and support decision-making in dynamic and unstructured environments. I am deeply interested in how these capabilities can allow robots to better understand their surroundings and make informed decisions in real-world settings.

Research & Development Experience

At the Autonomous & Undersea Systems Division of Aerospace Engineers Private Limited (AEPL), I led the research and development of electrical and embedded systems for unmanned marine vehicles, including Autonomous Underwater Vehicles (AUVs), Remotely Operated Vehicles (ROVs), and Autonomous Surface Vessels (ASVs). I worked across the complete electrical and embedded development cycle, from schematic and multilayer PCB design to board bring-up, power distribution, propulsion electronics, sensor and payload integration, and final hardware integration. A major part of this experience was the development of the Neerakshi AUV, where I contributed to improving the electrical architecture, thermal management, propulsion system, and overall vehicle integration.

Semiconductor Capacitance Extraction Tool Thumbnail

Semiconductor Capacitance Extraction & Linearization Tool

Problem Statement and Solution:

In a power electronics research lab at Arizona State University, researchers needed to extract voltage-dependent capacitance characteristics from semiconductor datasheet plots for their research. Manual curve extraction takes time and can introduce errors, especially with nonlinear curves, logarithmic axes, and overlapping characteristics. To address this problem, I developed a human-in-the-loop computer vision pipeline combining interactive graph calibration, centroid-refined seed selection, and 8-connected BFS pathfinding to trace target curves. The tool converts extracted datasheet curves into usable capacitance data for further engineering analysis.

EdgeVision Thumbnail

EdgeVision: User Defined Object-Counting using Raspberry Pi

Problem Statement and Solution:

This project develops a real-time object detection and tracking system on a Raspberry Pi for monitoring and counting objects in video streams. A key challenge was automatically determining when a tracked object crosses a specific region of the video without relying on a fixed counting area. To address this, the system combines real-time object detection, centroid-based tracking, and geometric line-crossing logic. Users can draw line segments directly on the video feed, and the system detects when tracked objects cross these lines and automatically updates the object count.

SLAM Thumbnail

Comparative Analysis of SLAM Algorithms

Problem Statement and Solution:

This project compares GMapping and Cartographer SLAM for indoor mapping using a Yahboom ROS Master X3 robot equipped with an RPLIDAR. Mapping accuracy was evaluated by comparing room dimensions with dimensions extracted from occupancy-grid maps using quantitative error analysis. GMapping performed better in simpler, structured environments, while Cartographer performed better in complex layouts. Testing also identified blind spots caused by the 2D LIDAR's mounting height, leading to a proposed depth-camera sensor-fusion approach for detecting low-lying obstacles.

PCB Project Thumbnail

PCB for Manual and Autonomous Robots

Problem Statement and Solution:

Team Vector, our college robotics club, faced a wiring challenge during our participation in ABU Robocon-India. Connecting multiple electronic subsystems required a large number of individual wires, which led to loose connections, wiring confusion, and EMI/EMC issues during electrical rack assembly and debugging. To address this, I designed a custom PCB that consolidated the required functionalities and commonly used interfaces into a single board, simplifying the electrical connections and making the system easier to assemble and debug.

Battery Fuel Gauge Thumbnail

Battery Fuel Gauge

Problem Statement and Solution:

Team Vector's testing team had to remove the robot's electrical rack each time they needed to check the health of the battery pack using a battery checker. This made the process repetitive and cumbersome during testing. To address this, I developed a "Battery Fuel Gauge" system that provided an onboard indication of the battery pack's health through an LED array. The system also generated an alert when the battery charge dropped below a defined threshold, allowing the team to monitor the battery without repeatedly removing the electrical rack.

LCD & Relay Board Thumbnail

LCD & Relay Breakout Board

Problem Statement and Solution:

After integrating the "Battery Fuel Gauge" system, Team Vector wanted to monitor the battery pack status without accessing the robot's internal electronics. To address this, I developed an external LCD display that provided a clear view of the battery status from outside the robot. The display was also designed for other system-status and control applications, while the breakout board included relays to support additional switching and control functions. This made the board a flexible interface for both battery monitoring and other robot subsystems.

PCB Ruler Thumbnail

Multipurpose PCB Ruler

Problem Statement and Solution:

During PCB design and debugging, I frequently needed to look up PCB-related calculations and component footprints online. This interrupted the design process and made it difficult to quickly access commonly used information. To address this, I developed a multipurpose "PCB Ruler" that integrates essential PCB calculations and commonly used component footprints into a single reference tool. This allowed other designers and testers to access frequently needed information directly during board development and debugging.

Line Follower PCB Thumbnail

PCB for Line Follower Robot

Problem Statement and Solution:

When Team Vector, our college's robotics club, welcomed new members, we organized a workshop built around a challenge involving obstacle avoidance, path following, and flag retrieval. The Electronics department was responsible for designing and fabricating a PCB in-house using the limited components available. As the trainer, I guided the new members through the complete process, from PCB design to board bring-up, giving them hands-on experience while emphasizing practical problem-solving and teamwork.

Electric Scooter Thumbnail

Retro-converted Electric Scooter

Problem Statement and Solution:

A friend of mine owned a petrol-powered Honda Aviator that he was considering selling because he was no longer using it due to its reduced mileage. We saw an opportunity to convert it into a fully electric scooter. While searching for suitable components in the local market, we came across an abandoned electric two-wheeler that provided several usable components for the conversion. This discovery gave us the opportunity to take on the project and convert the Honda Aviator from a petrol-powered scooter into an electric one.

Team Nikola Thumbnail

Team Nikola Racing

A team of 20 students from K. K. Wagh Institute of Engineering Education & Research, Nashik, came together to design and build an electric bike as a collaborative EV project. The team worked across the design, manufacturing, and fabrication of the vehicle. As part of the core electrical team, my friend and I worked on the vehicle's electrical system, including battery-pack development, BMS integration, and charging. The bike achieved a charging time of approximately 20 minutes and a battery lifespan of more than 8 years.

Team Vector Thumbnail

Team Vector

Team Vector is a robotics club comprising 40 students from different engineering disciplines at K. K. Wagh Institute of Engineering Education & Research, Nashik. The team brings together students with diverse technical skills and experience in robotics, working collaboratively across different areas of robot design and development. Team Vector has participated in ABU Robocon-India since 2016, building its technical skills, experience, and competitive performance over the years.

Get in Touch

I'm always open to conversations, collaborations, and exploring opportunities. Whether you're looking to connect for internships, full-time positions, or just to discuss innovative ideas, reach out. I'm just a message away!

  • Location

    Tempe, AZ

  • Email

  • Professional Links & Resume