Learn The Hidden Secret of Autonomous Car Navigation & Guidance: IMUs

Learn The Hidden Secret of Autonomous Car Navigation & Guidance: IMUs

Sensors Insights by Dan Dempsey

Inertial sensors - accelerometers and rate gyroscopes - are the “inner ear” of the car and have been in cars for many years doing basic, low performance tasks in applications like airbags and stability control systems. 

Without input from any other sensors, inertial sensors can detect the movement of the vehicle. For example, a simple accelerometer can detect the high-g sharp deceleration of a car to deploy the airbag.  A slightly more advanced inertial sensor package consisting of two orthogonal XY accelerometers and a single or dual-axis rate gyroscope is often used for vehicle stability control.  In vehicle stability systems lateral and longitudinal accelerations as well as rotation rates determine if the car needs to take some action to prevent roll over or minimize wheel slippage through a turn.  Essentially the inertial sensor package determines if a car is moving in a way that does not agree with driver input. 

While existing vehicle-mount inertial sensor package measure one of more axes of motion, an Inertial Measurement Unit or IMU is a module embeds a 3-axis linear accelerometer and 3-axis rate gyroscope to measure the full six degrees of freedom ("6 DOF or 6-axis").  Owing to the 6-axis configuration with both linear (X Y Z) and rotation components (roll, pitch, and yaw), the IMU captures all components of vehicle motion.   An IMU can be used for more than air-bags and vehicle stability; an IMU can track full vehicle position and orientation in real-time.   Specifically, when combined with an Extended Kalman Filter and calibrated to eliminate temperature and bias drifts, the IMU can very accurately determine the position of a vehicle for short periods of time without any assistance.  More advanced systems sometimes add in wheel speed and wheel angle to assist in the position estimations by the Kalman filter - further improving accuracy.


Sensors 2019 Hits Silicon Valley June 25-27!

Join 7,000+ engineers this June in San Jose at the sensor industry’s biggest event! With 65+ Technical Sessions, 100+ Leading Speakers, 10 new and updated tracks, and 300+ Interactive Exhibits, there’s more opportunity than ever to connect with this booming industry and the technologies driving it. See thousands of the newest technologies in action, learn about the latest applications, and develop invaluable partnerships at the only event dedicated to sensors, connectivity, and systems.
Fig 1: Next generation autonomous vehicles will need to combine motion sensors (IMUs) with location sensing (GNSS/GPS receivers) to safely operate.
Fig 1: Next generation autonomous vehicles will need to combine motion sensors (IMUs) with location sensing (GNSS/GPS receivers) to safely operate.

The latest generation of Advanced Driver Assistance Systems (ADAS) and autonomous vehicles require a very precise IMU to accurately predict the motion of a vehicle to determine its exact position in real-time.  In these advanced systems, the IMU information is combined with GPS receivers and even vision sensors such as Lidar and Cameras to continuously estimate the vehicle’s position and feed that information to the system’s central computing module   Navigation that relies on an IMU fused with additional sensor data such as GPS is often referred to as an Inertial Navigation System or INS.

By itself, a GPS receiver cannot provide highly accurate continuous position information.  With optimal reception of a few dozen global satellites the GPS receiver can calculate its position to within a few meters.  By also including correction signals to correct satellite clock errors and atmospheric distortion the GPS receiver can calculate position to around 2 cm to 4 cm using algorithmic techniques such as Real-Time Kinematics (RTK). The calculations take time to perform so the update rate for the GPS receivers is typically about 1 Hz or once a second but can be as fast as 10 Hz to 20 Hz for more dynamic applications.   At highway speeds, the vehicle can move 10 ft. or so between GPS updates under optimal conditions.

It is the role of the IMU to estimate the vehicle’s position between each update of the GNSS/GPS receiver.  Additionally, as GPS receivers frequently lose signal in “GPS challenged” environments such as in tunnels and near tall buildings, the IMU needs to be able to estimate position for 10, 20 and even 30 seconds.  Due to the required precision of the self-driving system, the allowable drift needs to stay in the 10 to 30 cm range.  While military and research grade equipment can today deliver this performance, the cost of these systems could exceed the cost of today’s entry level car models. A typical military grade IMU can cost in the five figures!

Instead, to deliver reasonably priced IMUs, developers today often use MEMS based accelerometers and gyroscopes.  Silicon based MEMS sensors are manufactured in high volumes thereby ensuring that their price is more suited to consumer and industrial based systems, with prices for a well calibrated unit under a hundred dollars each in quantity!  The next generation of MEMS IMU sensors promise to offer the accuracy and reliability required for advanced automotive applications up to and including totally automated L5 level driving applications.

MEMs based inertial measurement units offer the promise of size and process suitable for the automotive market.  There are several good performing MEMS IMU sensors on the market with Gyroscope Bias Instability (BI) of 5o/hr., Angle Random Walk (ARW) of 0.5 o/√hr. and Acceleration BI in the 10µg range.  These products can certainly provide effective position information and smooth out the intervals between GPS updates.  However, as vehicles move through tunnels and underpasses these intermediate grade IMUs will struggle to maintain position accuracy of less than 10cm after just a few seconds.  The current state-of-the-art MEMS inertial sensors are working to deliver Gyro BI close to 1o/hr. with ARW of 0.1 o/√hr.  If solutions with this level of performance can be adopted, integrated Inertial Navigation Systems (GPS+IMU) will be able to deliver the performance needed for advanced autonomy.

Fig 2: The compact ACEINNA OpenIMU family provides engineers with a high level of flexibility regarding where and how to integrate the guidance component within a vehicle.
Fig 2: The compact ACEINNA OpenIMU family provides engineers with a high level of
flexibility regarding where and how to integrate the guidance component within a vehicle.

In addition to the above issues, MEMs sensors have an inherent stiction issue in which the microscopic silicon fingers of the accelerometer and gyroscope structure can get stuck together.  Usually caused by a shock or high acceleration event, once stuck together they can be very difficult to separate them due to a phenomenon known as the Van der Waals force.  The devices cannot be power cycled like other semiconductor devices to correct the issue. 

The MEMS gyro/accel failure rate is acceptable in the consumer and industrial markets, but can it meet the low failure rate and long lifetime needs automotive companies demand?  This will be another challenge to the IMU developers.

Everyone is looking forward to the day when truly autonomous vehicles can replace our current antiquated vehicles and free up valuable space in garages and parking lots, while making our roads safer and much more efficient. Worldwide, thousands of engineers are working to develop the next generation of sensing technologies to enable this evolving milestone in transportation.

Often unheralded, the basic technologies of IMU and INS navigation are essential elements to ensure safe and efficient autonomous vehicles.


About the author

Dan Dempsey is the Senior Director of Automotive Business Development for ACEINNA.  He is responsible for expanding the inertial-navigation solutions into the automotive market.  Prior to ACEINNA, Dan Dempsey worked at Maxim Integrated for over 10 years.  Most recently as the Managing Director of Product Management and Product Marketing for several automotive analog and mixed signal product lines.  Dan has a BS in Biology from UCLA and an MSEE from Stanford.

Suggested Articles

From sensor to cloud, learn the potential pitfalls threatening signal purity.

Sensor detects level via vibration technology.

Pressure gauge suits applications that do not require all stainless steel.