AURORA – Autonomous Mobile Robot (AMR) - Currently under development

AURORA is a state-of-the-art Autonomous Mobile Robot (AMR) developed by our team to address real-world logistics, industrial transport, and indoor mobility challenges. Built with modularity, intelligence, and adaptability in mind, AURORA integrates high-performance hardware with an intelligent software stack capable of localization, navigation, environment sensing, and fail-safe autonomous operations.

Overview

Project AURORA is an intelligent autonomous ground robot engineered for logistics, research, and facility automation.
It merges robust mechanical design with dual-layer computing and AI-driven autonomy — enabling it to carry heavy payloads, navigate dynamic spaces, and adapt to changing environments through SLAM, sensor fusion, and predictive control.

“AURORA isn’t just an AMR — it’s an adaptive machine that learns and reacts.”

Objectives

Develop a payload-ready, modular 4WD robot with adaptive power management.
Implement SLAM-based navigation and local obstacle avoidance.
Achieve reliable multi-hour endurance under full load.
Enable fleet operation with decentralized communication and smart docking.

System Architecture

Layer	Description
Mechanical Layer	Aerospace-grade anodized aluminum frame, 4WD skid-steer, modular deck for flexible payloads.
Energy Layer	48 V 30 Ah Li-ion smart pack with dual-tier regulation and regenerative braking.
Compute Layer	Dual-tier stack – Raspberry Pi 5 (vision + SLAM) + Arduino Mega 2560 (real-time control).
Sensor Layer	LIDAR 270°, ToF arrays, IMU, accelerometers, gas/temp sensors.
Control & Comm.	Wi-Fi / BLE interface, PID motion control, ROS 2 middleware for autonomy.

Mechanical & Electrical Highlights

4WD skid-steer drive with independent torque distribution.
Payload: 100 kg tested (8 incline trials < 2.5° deviation).
Combined weight: ~140 kg.
Battery hot-swap supported; runtime 6–8 hours.
Max speed: ~1.5 m/s.
Thermal stability: < 45 °C after 6 hr full-load run.

Control & Autonomy Stack

Localized SLAM with Recovery Mesh: Retains map memory across layout changes.
Priority Path Planner: Selects fastest or lowest-energy route per payload.
Predictive Obstacle Avoidance: LIDAR + ToF fusion anticipates occlusions.
Micro-PID Adaptation: Adjusts torque response to uneven load.
Failover Redundancy: Auto-isolates faulty modules during runtime.

Performance Metrics

Metric	Result
Payload Stress	100 kg / 8 incline trials (< 2.5° deviation)
Endurance	7.4 hr runtime @ 83% avg draw
Path Accuracy	> 95.3% (indoor randomized layout)
Obstacle Response	~ 0.7 s average delay
Thermal Control	< 45 °C for 6 hr continuous run
Recovery	3/3 auto-stabilizations post fault

Challenges & Solutions

Challenge	Response / Technique
IMU drift in magnetic zones	Multi-instance averaging + external orientation calibration
Voltage sag under torque spikes	Capacitor buffer + torque-ramp sequencing
LIDAR occlusion near ground	Redundant ToF scanning + tilting base
Sensor-loop overlap	Mutex-locked cycle segmentation

Key Subsystems

Subsystem	Controller	Function
Drive & IMU	Arduino Mega 2560	PID control, encoder feedback
Vision & SLAM	Raspberry Pi 5	LIDAR processing, path planning
Power	Smart BMS	Voltage / current monitoring + recovery
HMI	OLED + LED Bar + Buzzer	Real-time feedback & status
Communication	Wi-Fi / BLE / Ethernet	Telemetry + fleet coordination

Testing Phases

Phase	Objective	Target Outcome
1	CAD validation + simulation (done)	Control abstraction verified
2	Frame + power integration	Electrical safety & routing done
3	Drive + sensor calibration	Stable motion + base mapping
4	SLAM trials + payload tests	95% path accuracy achieved
5	Endurance + failover tests	Multi-hour runtime validated

References

ISO 3691-4:2020 – Autonomous Truck Safety Standards
Raspberry Pi 5 SoC Developer Documentation
Arduino Mega 2560 Electrical Datasheet
ROS 2 Localization and Path Planning Docs
IEEE Xplore – AMR Deployment Models for Structured Environments