How MEMS inertial sensors can rival GPS

Modern navigation was revolutionized by satellite-based global positioning system (GPS) technology, but it’s not always available. GPS relies on signals from space, so it’s susceptible to anything that disrupts that journey — be it tall buildings, the deep ocean, or electronic interference.

To address this challenge, HRL Laboratories, an IBM company, has developed a new type of inertial measurement unit (IMU) that employs an advanced algorithm to provide an uninterruptible alternative to GPS.

“We're getting to the point where over-reliance on GPS has actually become a problem,” said David Chang, who leads the team working on HRL’s IMUs. “Inertial sensors help you deal with the scenarios where you cannot rely on GPS.”

The heart of the IMU navigation approach is a fully self-contained device that precisely measures a vehicle’s movement in three dimensions. If you know where you started, and you know the exact path you’ve taken based on speed, orientation, and elapsed time, then you can work out your current coordinates. This approach revives a pre-GPS navigation technique called “dead reckoning” which is how some craft, such as submarines, still find their way around.

But while a submarine’s high-performance IMU can be about the size of a dorm fridge, HRL’s IMU uses inertial sensors built using micro-electromechanical systems (MEMS) to yield a device about the size of a soup can. This small form factor makes it versatile enough to be deployed on lightweight vehicles like aerial drones.

How do MEMS inertial sensors work?

MEMS inertial sensors use miniaturized versions of accelerometers and gyroscopes etched into processor-sized silicon chips to track changes in velocity and angle of travel.

Accelerometers exploit a suspended weight’s resistance to changes in motion — its inertia — to compute the difference in velocity between it and the rest of the device. Imagine holding a cup of water in a car. If the vehicle quickly comes to a stop, the water will keep moving forward and end up splashing out of the cup. The suspended weights in MEMS accelerometers will similarly lag behind as a vehicle speeds up or swing forward as the vehicle slows down, triggering a signal in the surrounding sensors that corresponds to the acceleration being applied.

MEMS gyroscopes also measure movement based on inertia, but in their case, it’s rotation. The MEMS sensor contains a mass that vibrates back and forth. As the gyroscope spins, the inertia of the mass causes it to experience a shift in movement perpendicular to the direction it’s vibrating. Sensors that surround the mass pick up on this sideways motion to calculate the angular velocity causing it.

A high-performance MEMS inertial sensor is designed to measure changes in one dimension to avoid unwanted cross-axis sensitivity. To account for this, the team’s IMUs contain three of each MEMS sensor aligned to the three-axis coordinate system describing the physical world.

Versions of MEMS accelerometers and gyroscopes are found in many consumer devices — this is how a fitness tracker can tell whether you had a restless night of sleep and how a phone knows to spin its screen around when it gets flipped — but HRL’s MEMS inertial sensors are engineered to be much more accurate in order to come close to GPS for navigational accuracy.

Solving bias drift is essential for a true GPS alternative

For a long time, a phenomenon called “bias drift” has been limiting the use of MEMS-based IMUs for navigation. Bias drift occurs when noise from sources like vibrations and heat throw off sensors and generate a false inertial signal — in other words, bias is the signal from a sensor when it is not moving. The longer a typical MEMS-based IMU is used, the more bias drift compounds, undermining the ability to navigate for longer than a handful of minutes.

But now, bias drift is an issue that HRL researchers have evidently solved.

“The holy grail for these MEMS sensors is to deal with drift. You need to figure out how to calibrate it out,” said Chang. “I think we’ve finally cracked that nut.”

The team’s solution is AutoCal, an automatic, self-calibrating system that continuously measures bias and corrects for its drift. The system works by regularly taking two simultaneous measurements for each sensor in opposite directions, capturing inertial signals with equal magnitude but different signs. Since the bias is the same nondirectional quantity contained in both measurements, it’s possible to arithmetically isolate the bias while also retaining the true core signal for navigation. The bias is then removed from subsequent measurements until the next calibration.

Continuously correcting for drift means that these IMUs can be used indefinitely to support navigation, removing a major barrier to matching the accuracy of GPS.

How are MEMS-based IMUs optimized for cost and size?

In addition to solving for bias drift, the researchers want to make high-performance IMUs small and cheap enough to deploy widely.

One approach to saving space is using application-specific integrated circuits (ASICs) for custom electronics that reduce the bulk dedicated to off-the-shelf parts. Advanced packaging techniques were also developed to minimize the amount of unused space between components. This involves considerations like how the MEMS devices are assembled and the design of signal connections. Together, these efficiencies add up to a significantly leaner device.

To keep costs down, the research team ensured its MEMS chips — which have the added challenge of moving parts like springs — could be mass-produced on silicon wafers with semiconductor manufacturing techniques, including using tools for etching precise patterns and depositing materials in exacting quantities. Partnering with established fabrication facilities provides a pathway to benefit from the same economies of scale as producing computer processors.

What’s next for MEMS inertial sensors?

The team aims to keep shrinking its AutoCal IMUs. The next step is to get the device down to about the size of a hockey puck.

HRL is also looking to expand the applications that IMUs can support. While navigation is the biggest opportunity on the horizon, any use case that employs GPS could eventually benefit from MEMS-based IMUs. For example, they could help autonomous vehicles get around cities where buildings often disrupt satellite signals.

But the researchers’ immediate focus is getting their current IMU into the hands of real-world users, with production expected to begin in 2027. “Cost, size, weight, and power — these factors are very important if you want to proliferate these devices,” said Chang. “I think at some point we're going to run into a situation where GPS is just totally unreliable for critical applications.”