HARDWARE SPECIFICATION

The Technology Stack

Resku is built entirely on accessible, commodity hardware. Every component is available off the shelf. Every line of software is open source.

Compute Core

The Raspberry Pi 4 Model B serves as the central processing unit — running the SLAM engine, managing sensor I/O, and maintaining the live spatial map in real time.

Component Specification Role
Raspberry Pi 4 Model BBroadcom BCM2711 · 4-core 1.8 GHzMain compute · SLAM runtime
RAM4 GB LPDDR4-3200Map buffer · process isolation
Storage32 GB Class 10 microSDOS · map persistence
OSRaspberry Pi OS Lite (64-bit)Headless runtime
PowerUSB-C 5V/3A · optional LiPo HATField portable

Sensing Array

Three complementary sensors on the I²C bus and a dedicated UART LiDAR unit provide full spatial and thermal situational awareness.

Sensor Interface Key Spec Purpose
RPLIDAR A1M8UART / USB360° · 5.5 Hz · 12 m rangePrimary LiDAR — 2D SLAM
MLX90640I²C (0x33)32×24 px · −40 to +300°CThermal array — victim detection
MPU-6050I²C (0x68)6-DOF IMU · 16-bit ADCGyroscope + accelerometer — odometry correction
BMP280I²C (0x76)±1 hPa · ±0.5°CAmbient temp / pressure logging

Software Stack

The SLAM pipeline is implemented in Python, leveraging established robotics libraries with a lightweight custom mapping layer tuned for resource-constrained hardware.

Component Library / Tool Notes
SLAM EngineHector SLAM / custom EKF2D occupancy grid mapping
LiDAR Driverrplidar-pythonSlamtec SDK wrapper
Thermal Driversmbus2 · custom MLX90640 lib32×24 frame @ 4 Hz
IMU Fusionmpu6050-raspberrypiComplementary filter
Map OutputPGM / PNG via PillowCompatible with ROS map_server
RuntimePython 3.11 · asyncioNon-blocking sensor loop

How SLAM Works

Simultaneous Localization and Mapping (SLAM) allows the module to build a spatial map of its surroundings while simultaneously tracking its own position within that map — without GPS or any prior knowledge of the environment.

STEP 01

Scan

LiDAR emits 360° laser pulses, measuring the time-of-flight to reflect distance to every surface within range.

STEP 02

Match

Each new scan is compared against the current map using iterative closest point (ICP) matching to estimate movement.

STEP 03

Update

The occupancy grid is updated with new obstacle data, and the EKF corrects positional drift using IMU sensor fusion.