Building Safer Systems with Multi-Sensor Integration

Article by: Mane Harutyunyan (DEPI Space: Physics Research Engineer)

Creator: David Mayes KSU DCM Photographic Services | Copyright: ©Kansas State University

Abstract In safety-critical engineering, multi-sensor integration is the bedrock of reliability and resilience. Few examples are as compelling as the life-support system in an astronaut’s suit. This isn’t just about collecting data; it’s about weaving a safety net so intelligent that it can predict failure before it happens. By prototyping a simplified version of this system, we can uncover the powerful principles that protect lives, both in space and here on Earth.

Introduction Imagine the silence of space, broken only by your own breath inside a helmet. Your survival depends entirely on the machine on your back — the Portable Life Support System (PLSS). It’s a miniature spacecraft, responsible for your air, your temperature, and your pressure.

Modern systems, from autonomous cars to smart hospitals, face similar (if less dramatic) stakes. They all depend on distributed sensor networks to make life-or-death decisions. The key lesson from aerospace is simple: trusting a single sensor is a gamble you can’t afford. True safety emerges from the fusion of multiple sensing modalities, creating a system that is redundant, cross-validated, and context-aware.

The Power of Multi-Sensor Integration:

The PLSS is a masterclass in closed-loop control. It doesn’t just monitor; it acts. This real-time responsiveness is only possible through sophisticated sensor integration, which enables:

• Redundancy: Multiple sensors for critical parameters (like oxygen) ensure a single faulty reading doesn’t become a fatal event.
• Cross-verification: Data fusion algorithms constantly check sensors against each other. A pressure sensor reading a drop isn’t an immediate alarm; it’s a hypothesis that needs confirmation from a CO₂ or temperature sensor.
• Contextual Awareness: A motion sensor (gyroscope) showing an astronaut is highly active provides context for a slightly elevated heart rate or CO₂ output. Without that context, the system might misinterpret healthy exertion as a system failure.
• Proactive Response: This integrated data stream allows the system to not just sound an alarm, but to take corrective action — like adjusting coolant flow or initiating a backup oxygen supply.

From Theory to Practice

To move beyond theory, we built a functional desktop prototype simulating core PLSS functions. The goal wasn’t to build flight-ready hardware, but to tangibly explore the challenges and payoffs of sensor fusion.

Our system integrated five key sensors:

Sensors

BMP180 – Suit Pressure Monitor
This sensor keeps track of the pressure inside the spacesuit. 
If it suddenly drops, that could mean a dangerous leak. 
Even a slow change might signal a valve issue—both are critical to catch early.

MPU6050 – Motion Tracker
By measuring acceleration and rotation, this sensor shows whether the 
astronaut is moving around a lot or staying still. 
That context helps mission control understand body signals, 
like whether high heart rate comes from stress or just physical effort.

LM35 – Temperature Sensor
The LM35 keeps an eye on the astronaut’s temperature. 
It makes sure they don’t overheat from activity or get too 
cold in the shadows—comfort and survival both depend on it.

SEN0322 – Oxygen Monitor
This sensor checks the oxygen concentration in the breathing loop. 
Oxygen has to stay within strict limits: too little leads to dizziness 
and blackouts, too much increases the risk of fire.

MH-Z19C – Carbon Dioxide Detector
Since CO₂ is the main waste product of breathing, 
this sensor ensures it doesn’t build up inside the suit. 
A rise in CO₂ levels is dangerous and usually means the scrubbing 
system isn’t working properly.

Figure 1. Experimental setup for testing MPU6050, BMP180, LM35, SEN0322, and MH-Z19C sensors using Arduino Mega | Creator: Mane Harutyunyan | Copyright: ©DEPI Space

The Control System’s Lifeline

Collecting this data isn’t an academic exercise; it’s the central nervous system for a robotic caretaker. Each data point feeds a control algorithm that makes decisions.

For example:

1. The Data: The CO₂ sensor detects a concentration rising toward a dangerous threshold.
2. The Cross-Check: The system checks the motion sensor. Is the astronaut exerting themselves? (This would naturally produce more CO₂). The motion data is high.
3. The Context-Aware Decision: Instead of triggering a major CO₂ scrubber alarm, the system simply increases the fan speed slightly to handle the increased load. It logs the event as “expected metabolic increase.”
4. The Alternative: If the CO₂ was rising while the motion sensor showed the astronaut was static, it would trigger a high-priority alert for a possible scrubber failure.

Related Work and Comparison with Our Research

Several studies in the literature examine sensor technologies and their applications in health monitoring and portable support systems. These works provide valuable insights into innovative sensing methods, integration strategies, and reliability analysis. However, their scope and testing environments differ significantly from our focus on Portable Life Support System (PLSS) sensor testing for aerospace applications.

For instance, Portable Life Support System: Current Status and Future Perspectives outlines the general design, ergonomics, and integration challenges of PLSS devices but does not focus on detailed sensor testing. Our work addresses this gap by experimentally evaluating individual sensor performance within a PLSS context.

The article Sensor Systems for Prognostics and Health Management emphasizes theoretical frameworks for reliability assessment and lifetime prediction of sensors. In contrast, our research is hardware-oriented, concentrating on electrical characteristics, power consumption, and practical validation. Together, these approaches can complement one another — analytical models supporting experimental findings.

In summary, while prior studies contribute conceptual frameworks, prognostic models, and novel sensing methods, our research is distinguished by its hands-on experimental testing and mission-relevant evaluation of PLSS sensors, bridging the gap between theory and practical application.

Confronting the Real-World Challenges

Our simple prototype immediately highlighted the engineering hurdles that make systems like the real PLSS so remarkable:

1. Calibration Drift: Sensors “age.” An oxygen sensor might read 20.9% today but 20.5% next month. Without constant software-driven calibration against a baseline, this drift leads to catastrophic errors.
2. Temporal Alignment: If the motion data is a half-second behind the CO₂ data, the control system completely misinterprets their relationship. Precise time-synchronization between all sensors is non-negotiable.
3. The Power Paradox: More sensors and faster processing drain batteries. Designing these systems is a constant trade-off between analytical depth and energy efficiency.

Conclusion: The PLSS teaches us that safety isn’t a component you can bolt on; it’s an emergent property of a well-integrated, intelligent system. By prototyping with pressure, motion, and gas sensors, we felt the real friction of data fusion — and witnessed its incredible power.

These principles are already descending from orbit. Wearable health monitors are becoming proactive guardians, industrial sensors are predicting equipment failures, and smart cars are building a 360-degree view of the world to avoid accidents.

Multi-sensor integration is the foundation upon which we will build the next generation of intelligent systems that don’t just operate, but truly protect.

References

[1] Shunfeng Cheng, Michael H. Azarian, Michael G. Pecht, ‘’ Sensor Systems for Prognostics and Health Management’’, 2010

[2] Jingang Jiang, Yihao Chen, Xuefeng Ma, Yongde Zhang, Zhiyuan Huang, Dexian Liang, ‘’ Portable Life Support System: Current Status and Future Perspectives’’, 2010