Docs
Technical documentation and system design notes.
Longform documentation with architecture, simulation methodology, and experimental results.
Introduction
AmphiReach (AR) is an amphibious autonomous rescue robot built for long-range open-water casualty retrieval and shore-to-medic transfer. The system is designed to reduce time-to-first-contact, stabilize the casualty, and deliver a consistent handoff to medical teams.
Key Idea
AR closes the gap between flotation delivery and a complete rescue workflow by integrating navigation, gentle recovery, and amphibious transfer.
Scope
- Long-range open-water missions
- Gentle recovery pod with thermal guidance
- Amphibious transition for shore delivery
Design goals
- Reduce time-to-first-contact without exhausting energy reserves.
- Maintain casualty safety from recovery through handoff.
- Preserve operator oversight while enabling autonomous decisions.
Requirements
The system requirements reflect safety, autonomy, and environmental constraints.
| Requirement | Target |
|---|---|
| Range | 40 km nominal cruise |
| Sea state | Operable up to Beaufort 5 |
| Payload | 100 kg casualty payload |
| Detection | 90%+ success in low light |
| Transfer | Shore handoff within 10 minutes of arrival |
System Overview
AR is composed of a sensing mast, an amphibious hull, a recovery pod, and a modular compute stack. The autonomy pipeline connects mission planning with perception-driven recovery and shore transfer.
Subsystems
- Perception: thermal + RGB sensing and tracking
- Localization: GPS, IMU, and drift models
- Planning: multi-objective routing for time and energy
- Control: adaptive thrust and heading control
- Morphing: amphibious drivetrain transition
Mechanical Design
The hull is optimized for stability in moderate sea states with a sealed recovery pod integrated into the centerline. The drivetrain supports water propulsion and tracked shore mobility.
Hull and pod
- Low-drag profile for long-range cruising
- Passive flotation reserves around the pod
- Heated enclosure with drainage channels
Hardware
AR uses a ruggedized compute stack with redundant power rails, a thermal camera, and a stabilized gimbal. Modular hardware makes maintenance and upgrades predictable.
export const sensors = [
"thermal-camera",
"rgb-camera",
"imu",
"gps",
"anemometer"
];Software
The software stack uses modular autonomy services with clear interfaces between perception, planning, and control.
Services
- Mission coordinator
- Perception inference pipeline
- Planning and safety supervisor
- Actuation controller
Perception & Localization
Thermal detection provides robust candidate extraction in low-visibility conditions. Localization fuses GPS, IMU, and water-current estimates for stable tracking.
Detection pipeline
Thermal segmentation → candidate filtering → track association → recovery alignment.
Planning & Control
Planning balances time, energy, and safety margins. Control maintains steady heading and pod alignment for safe retrieval.
const plan = {
profile: "fast",
cruiseSpeed: 2.6,
driftCompensation: true
};Simulation & Methods
Simulations include varying sea states, current profiles, and lighting conditions. Results emphasize time-to-first-contact and recovery stability.
Evaluation
- Route efficiency
- Detection success in low light
- Energy consumption profiles
Results
Preliminary results demonstrate improved time-to-first-contact and stable recovery in moderate conditions.
| Metric | Outcome |
|---|---|
| TTFC reduction | ~15% |
| Detection success | 89-96% |
| Path overhead | 9-18% |
Discussion
Results indicate strong performance in moderate sea states with opportunities to reduce energy overhead in aggressive profiles. The recovery sequence remains the most sensitive stage and continues to be refined.
Limitations
- Extreme sea states beyond Beaufort 5 reduce recovery stability.
- Thermal detection performance drops in heavy precipitation.
- Shore transfer requires clear ingress paths.
Future Work
Upcoming milestones include expanded sensor redundancy, field trials in mixed sea states, and a refined recovery pod suspension system.
References
- Maritime rescue robotics literature survey (internal)
- Autonomous surface vehicle navigation benchmarks
- Thermal perception evaluation notes