An ongoing Micromouse robotics project focused on developing high-speed autonomous maze-solving robots, from custom electronics and mechanical design to embedded firmware, motion control, state estimation, and path planning.
This repository documents the evolution of the platform across multiple generations, including Chia v1, Chia v2, and Blaze v1.
Micromouse is an autonomous robotics challenge where a small mobile robot must explore and solve a maze without human intervention.
This project focuses on developing a complete high-performance Micromouse platform capable of:
- Real-time maze exploration and mapping
- Flood-fill based path planning
- High-speed autonomous navigation
- Precise motion control
- Encoder-based velocity estimation
- Sensor-based wall detection
- State estimation using Kalman filtering
- PID feedback control
- Feedforward motor control
- Trapezoidal motion profiles
- Custom PCB and electronics design
- Embedded firmware development
The robot is designed to navigate a 16×16 maze while maintaining accurate position, velocity, and heading control.
The project has evolved through several generations, with each version introducing improvements in hardware, electronics, firmware, and motion control.
The first-generation Micromouse platform was developed around a Teensy 4.1 microcontroller.
- Teensy 4.1 microcontroller
- Custom electronics
- Embedded firmware
- Wall-detection sensors
- Encoder-based motor control
- Autonomous maze navigation
- Custom mechanical platform
Chia v1 established the initial hardware and software architecture for the Micromouse platform.
Chia v2 introduced a more compact and integrated electronics platform while improving the mechanical and electrical design.
- Teensy 4.1 microcontroller
- Custom-designed 2-layer PCB
- 1000 RPM DC motors
- Integrated motor drivers
- Sensor interfaces
- Power-management circuitry
- Improved mechanical design
- Embedded motion-control firmware
The Chia v2 PCB integrates the microcontroller interface, motor drivers, sensor interfaces, and power-management circuitry into a compact electronics platform.
Blaze v1 is the third-generation Micromouse platform, developed with a stronger focus on high-speed autonomous navigation and optimized motion control.
- STM32 microcontroller
- Custom-designed 4-layer PCB
- Lightweight mechanical redesign
- High-speed DC motors
- Advanced motion control
- Sensor fusion
- Motion profiling
- Optimized embedded firmware
- High-speed autonomous navigation
The platform is designed for competitive Micromouse applications, with emphasis on smooth, accurate, and high-speed movement through the maze.
Blaze v1's motion controller runs a closed-loop feedback/feedforward scheme on both the forward and rotational axes, tracking smooth (non-abrupt) speed profiles rather than stepping directly to a target velocity. Wheel encoders provide the primary velocity feedback; an onboard IMU is fused in to improve heading accuracy, and wall-sensor readings contribute a steering correction to keep the robot centered while traveling through the maze. No further implementation detail is shared publicly at this stage.
flowchart TD
SYSID["MATLAB System Identification"] --> LMODEL["Left Wheel Model"]
SYSID --> RMODEL["Right Wheel Model"]
MP["Motion Profile Generator"] --> MIX["Wheel Target Mixing"]
FWD["Forward Estimate<br/>Sensor Fusion"] --> MIX
ROT["Rotation Estimate<br/>Sensor Fusion"] --> MIX
MIX --> LTARGET["Left Wheel Target"]
MIX --> RTARGET["Right Wheel Target"]
LTARGET --> LMODEL
RTARGET --> RMODEL
LMODEL --> FF["Feedforward Controller"]
RMODEL --> FF
LTARGET --> FB["Feedback Controller"]
RTARGET --> FB
ENC_L["Left Wheel Encoder"] --> FWD
ENC_R["Right Wheel Encoder"] --> FWD
ENC_L --> ROT
ENC_R --> ROT
IMU["IMU (Gyro)"] --> ROT
WALL["Wall Sensors (IR)"] --> STEER["Wall-Following<br/>steering correction"]
STEER --> FB
WALL -.-> DIST["Friction & Sharp IR<br/>Disturbance Model<br/>(under construction)"]
DIST -.-> FF
FF --> SUM(("+"))
FB --> SUM
SUM --> MOTOR_L["Left Motor"]
SUM --> MOTOR_R["Right Motor"]
MOTOR_L -.-> ENC_L
MOTOR_R -.-> ENC_R
classDef model fill:#f5eefc,stroke:#7a3fa0,stroke-width:2px,color:#1a1a1a;
classDef sensor fill:#fff4e6,stroke:#c97a1a,stroke-width:2px,color:#1a1a1a;
classDef estimate fill:#eaf1fb,stroke:#2b4c7e,stroke-width:2px,color:#1a1a1a;
classDef controller fill:#e8f8ee,stroke:#1f7a4d,stroke-width:2px,color:#1a1a1a;
classDef motor fill:#fdeaea,stroke:#b23b3b,stroke-width:2px,color:#1a1a1a;
classDef wip fill:#f2f2f2,stroke:#999999,stroke-width:2px,stroke-dasharray:5 5,color:#555555;
class SYSID,LMODEL,RMODEL,MP model;
class ENC_L,ENC_R,IMU,WALL sensor;
class FWD,ROT,STEER,MIX,LTARGET,RTARGET estimate;
class FF,FB,SUM controller;
class MOTOR_L,MOTOR_R motor;
class DIST wip;
Due to project confidentiality before competitions, only selected PCB design details are publicly shared.
The robot uses a layered control architecture combining low-level motor control with higher-level navigation algorithms.
The robot continuously builds an internal representation of the maze using its wall sensors.
Flood-fill is used to calculate distances to the goal and determine efficient paths through the explored maze.
Motion profiles are used to control acceleration, constant-speed motion, and deceleration instead of applying abrupt velocity commands.
Closed-loop motor control uses encoder feedback to regulate wheel velocity.
The control system incorporates both:
- Feedback control
- Feedforward control
This allows the robot to achieve faster and more accurate velocity tracking.
Sensor measurements are combined using a Kalman filter to improve estimates of the robot's motion state.
MATLAB-based modelling and experimental measurements are used to characterize motor behaviour and tune the motion-control system.
| Generation | Microcontroller | PCB | Motor |
|---|---|---|---|
| Chia v1 | Teensy 4.1 | Custom electronics | DC motors |
| Chia v2 | Teensy 4.1 | Custom 2-layer PCB | 1000 RPM DC motors |
| Blaze v1 | STM32 | Custom 4-layer PCB | High-speed DC motors |
- C
- C++
- STM32
- Teensy 4.1
- Embedded firmware
- Encoder interfaces
- Motor control
- PID control
- Feedforward control
- Motion profiling
- State estimation
- Kalman filtering
- Differential-drive kinematics
- Maze mapping
- Flood-fill path planning
- Autonomous exploration
- Optimal path planning
- Altium Designer
- Custom PCB design
- Motor drivers
- Sensor interfaces
- Power management
- MATLAB
- System identification
- Experimental characterization
- Controller tuning
The Micromouse platform has been developed for autonomous robotics competitions, including Micromaze 2.0 at IIT, where the team was selected among the Top 10 finalists.
The system is continuously being improved toward faster exploration, more accurate motion control, and optimized maze-solving performance.