MEMS inertial sensors easily solve the "positioning" problem of emergency rescue

Therefore, first responder positioning remains the most complex positioning application today. While there is no silver bullet sensor that can achieve the desired goal, multiple technology nodes are required, each with leading-edge performance. In addition, it involves large-scale sensor fusion and system integration methods.

Cost-effective, high-performance MEMS inertial sensors can now provide the seeds for potential solutions. This article proposes a complete sensor-to-cloud sensor fusion system concept, including highly complex algorithms. Table 1 below describes the main methods and implementation technologies.

Table 1. Complete system approach to key objectives

The main challenges faced by system developers can be summarized into the following three categories: procedures, environment and sensor fusion. In the process of designing multi-sensor solutions, a comprehensive understanding of the high complexity of emergency missions and the challenges posed by various extreme environments is necessary.

Fire search and rescue missions must be carried out in strict accordance with rescue procedures, while at the same time having to adapt to completely uncertain real-life scenarios. A deployable precise positioning system must be adaptable to existing processes and equipment to the greatest extent possible. This requires the ability to operate without any fixed or temporary infrastructure, as first responders are often already carrying important equipment (weight and cost). Any system development should follow the early stage goal of achieving small embedded devices with a unit first responder cost similar to that of a smartphone. It is important to point out that the positioning performance of current smartphones is seriously insufficient, so challenges are faced. Figure 1 summarizes the most relevant primary and secondary operational requirements of the ideal system.

Figure 1. Key operational requirements define the first responder product design problem.

While GPS coverage makes outdoor positioning ubiquitous, it does not support fully indoor or mixed (indoor/challenging outdoor) environments. Some indoor positioning environments (such as shopping malls) can be achieved by installing infrastructure - however, these are neither accurate nor practical for emergency operations. For tracking system designers, the following factors must be considered to determine the design, component selection and risk reduction methods:

• RF propagation path.

• Sensor temperature/shock effects.

• Potential for damage/alteration of infrastructure.

The previously mentioned challenges in the process and environment are the basis for the core design approach to the sensor fusion problem. The relevant primary sensing modes are used to provide high performance in key operating modes, while complementary sensors remove key obstacles in each application stage, as shown in Table 2.

Table 2. Advantages and disadvantages of candidate sensors

Because MEMS require no external infrastructure and can provide precision sensing in dynamic environments, they can play a major role in the overall solution if they can operate in extreme environments and if used in conjunction with the right secondary sensors.

Consumer inertial MEMS devices have quickly moved toward commoditization (with a strong focus on performance specifications), military MEMS remain prohibitively expensive, and industrial and automotive MEMS (see Figure 2) are targeting both performance and cost levels.

Compared to the consumer sector, the industrial and automotive sectors require accurate sensing in relatively complex and extreme environments, and suppliers have integrated architectural features specifically to address factors that affect performance, such as off-axis motion, vibration and shock events, and errors caused by time and temperature. While these design features are often most easily accommodated through larger sensors or more expensive processing, the economic pressures of the automotive industry and the increasingly important industrial market are forcing a more critical approach to designing performance and achieving cost benefits.

Ultimately, cost-effective MEMS components were developed specifically for industrial applications, and the percentage of distance-related errors for the three main categories of components are compared in Table 3. Industrial-grade MEMS can provide navigation capabilities as good as high-end military equipment, while being reasonably priced compared to commodity consumer MEMS components.

Table 3. MEMS navigation performance level and transmission distance error percentage

The reason for this advantage requires a closer look at the key specifications of the MEMS component in relation to the target application. For the first responder operation target, a key task of the MEMS sensor is to identify the current type of movement and measure the number and length of steps. Unlike pedestrian movement models, first responder movement will be more random, dynamic and difficult to identify. In addition, due to the accuracy target, the sensor must be able to reject erroneous movements such as vibration, shock, and side-to-side shaking/swaying of the feet or body.

The first responder model is not a simple accuracy analysis of sensor noise, which may be sufficient for the pedestrian model, but must also include key specifications such as linear g rejection and cross-axis sensitivity. Figure 4 compares the three important RSS error specifications for industrial and low-end MEMS devices. It is easy to see that noise is not a detrimental factor, but linear g and cross-axis performance, which are not specified on many low-end devices, are the main issues.

Table 4. Comparison of RSS errors for industrial and low-end MEMS, showing that noise is not a factor in performance

Assumptions: 50 Hz BW, 2 g rms vibration, 100º/sec off-axis rotation.

Although just a few years ago, high-performance inertial sensors could only be achieved through methods such as optical fiber, now industrial MEMS processes have clearly proven that they are fully capable of doing so. A comparison of key navigation indicators is shown in Table 5 below.

Table 5. Comparison of key navigation indicators between cost-effective industrial MEMS and traditional fiber optic gyroscopes

An example of an industrial MEMS IMU is the ADIS16488A, shown in Figure 2, which incorporates 10-DOF high-performance sensing and is suitable for the most demanding applications, commercial avionics (as shown in Table 6), proving its readiness for emergency extreme applications.

Table 6. ADIS16488A MEMS IMU; cost-effective with proven high performance and reliability

Advances in inertial MEMS performance and continued proven quality and durability are now being combined with significant advances in integration. The last hurdle is particularly challenging because sensor size is inversely proportional to performance and durability if not carefully managed. A highly strategic, coordinated, and challenging series of process advances must be tested and merged to meet the performance density levels required for this application, as shown in Figure 3.

Figure 3. Industrial MEMS IMUs offer advanced performance, size, cost, and integration (without compromise) to support only critical applications such as first aid.

The selection of the appropriate sensors for a given application should be preceded by a thorough analysis to understand their weight (relevance) in the different phases of the overall mission. For PDR, the solution is driven primarily by the available equipment (e.g., embedded sensors in smartphones) rather than by performance design. As a result, there is a heavy reliance on GPS, with other available sensors, such as embedded inertial and magnetic, playing only a small role in determining useful position information. It works fine outdoors, but in challenging urban environments or indoors, where GPS is not available and the quality of other available sensors is poor, there are large gaps, or in other words, uncertainty in the quality of the position information. While advanced filters and algorithms are often used to merge these sensors without the need for any additional sensors or better quality sensors, software does little to close the uncertainty gap, ultimately only significantly reducing the confidence in the reported position. A conceptual illustration is shown in Figure 4.

Figure 4. Smartphone-based pedestrian navigation relies primarily on GPS, aided by non-optimized pre-embedded sensors, with large gaps in high confidence in motion detection or reliable coverage that cannot be fixed by algorithms alone.

In contrast, industrial dead reckoning scenarios, such as emergency operations, are designed for system-defined performance and component selection based on specific accuracy requirements. Better quality inertial sensors allow them to play a primary role, leveraging other sensors as appropriate to close the uncertainty gap. Rather than extrapolating/estimating position between reliable sensor readings, the algorithms are more conceptually focused on optimal weighting, switching, and sensor cross-correlation, as well as awareness of the environment and real-time motion dynamics (see Figure 5).

Figure 5. Sensors are specifically selected to fully cover the range of emergency missions, greatly improving the accuracy and reliability of the system.

Accuracy in either case can be improved by improving the quality of the sensor, and while sensor filtering and algorithms are an important part of the solution, they alone cannot eliminate the gaps in coverage of limited quality sensors.

In the specific case of first responder tracking, the mission was divided into the following phases to better assess the sensor processing requirements: arrival on scene, deployment, entry into the building and rescue - Table 7. Assuming that the fire truck is equipped with a high-end GPS/INS system, the position of the vehicle arriving on scene can be determined as a known reference point. From this point until the firefighter enters the building, there is an uncertain and random sequence of motion, and its precise location and mapping system relies on the ultra-wideband range implemented to accurately lock the firefighter's position and orientation. Once inside the building structure, inertial sensors become the primary tracking sensors, with the goal of providing a positioning accuracy of several meters.

If desired, the system can be designed to rely entirely on inertial sensors, but can also take advantage of other available and reliable random transmission signals, such as UWB range signals, magnetometer corrections, and barometric pressure measurements. As mentioned earlier, the implemented algorithm not only tracks position, but also generates a real-time path map for the search pattern. If a firefighter is down or in distress, the map generated by the initial path is supplemented by sensor input for the rescue firefighter, who is also guided by inertial sensing.

Table 7. Sensor requirements for different phases of emergency missions