I am working on a small indoor mobile robot designed to navigate around a house, workshop, or similar indoor environment without following a fixed, predefined path. The main goal of this project is to build an affordable robot that can understand its surroundings, identify obstacles, and determine a suitable route to reach a destination. I want to keep the overall design practical and relatively simple, using commonly available components instead of relying on expensive robotics equipment. A Raspberry Pi or similar single-board computer will serve as the main processing unit, handling mapping, navigation, and higher-level decision-making.
The mechanical design is based on a two-wheel differential-drive chassis with a small caster wheel positioned at the front or rear for balance and support. Two independently controlled drive motors will allow the robot to move forward, reverse, and turn by adjusting the speed and direction of each wheel. Similar principles of tracking movement and responding to changing conditions can also be seen in gaming, where players testing free fire beta rely on responsive controls and consistent in-game movement to evaluate gameplay performance. Wheel encoders will measure the rotation of the motors and provide information about how far each wheel has traveled. Using these measurements, the system can estimate the robot's movement and orientation through odometry. Although encoder-based odometry can accumulate errors over time due to wheel slip and uneven surfaces, it will provide an important starting point for estimating the robot's position.
For the sensing and mapping part, I plan to use a 2D LiDAR sensor to measure distances to walls, furniture, and other objects around the robot. As the robot moves, these measurements will help it recognize nearby obstacles and build a basic representation of the indoor environment. The Raspberry Pi will process the sensor data and combine it with movement estimates to support mapping and localization. This information will allow the robot to maintain an understanding of its surroundings instead of relying entirely on predefined movement instructions. The exact mapping and localization approach will depend on the available processing resources and the accuracy of the selected sensors.
The software architecture will separate high-level navigation from low-level motor control. The Raspberry Pi will handle tasks such as processing LiDAR scans, estimating the robot's position, building a map, and deciding which direction to move. A separate microcontroller will manage the motor controller and handle the time-sensitive tasks involved in driving the wheels. It will receive movement commands from the main computer and adjust motor outputs accordingly, using encoder feedback where appropriate. This separation should make the system easier to troubleshoot and maintain because navigation software and motor-control logic can be developed and tested independently.
Another important objective is to make the robot capable of moving safely through unfamiliar indoor spaces. Rather than following a fixed route, it should use its map and current sensor readings to select a suitable path and react when an obstacle blocks its way. For example, if a chair or box is placed in front of the robot, the navigation system should be able to recognize the obstruction and attempt to find another safe route when the surrounding space permits. I plan to develop the navigation functions incrementally, starting with reliable manual movement and encoder feedback, then adding LiDAR-based obstacle detection, mapping, localization, and autonomous navigation. Testing in different room layouts will help me understand how well the robot performs in practical conditions.
Overall, this project is an attempt to create a low-cost indoor autonomous robot that combines differential-drive mechanics, wheel odometry, LiDAR sensing, mapping, and intelligent navigation in one compact platform. I want the...
Read more »
Gaultier Lecaillon
Will Donaldson
jonathanr4242.utube
Keith Elliott
That's the right split. A few things that save time: ROS 2 with slam_toolbox on the Pi does the mapping well with an RPLidar A1, which is the cheap standard. Most diff-drive odometry error is heading, so add an IMU for yaw and calibrate the wheelbase with a square drive test before blaming the SLAM. Keep the motor PID on the MCU at a fixed rate and just send odometry to the Pi over serial. Disclosure: I run Compoden and we sell the A1 and a Pi 5 lidar mapping robot kit, if you'd rather not source it piece by piece.