The DriftLess⁺ technology is an improved version of the initial DriftLess technology invented & patented by TNO. This technology estimates and subsequent compensates for the offsets (biases) of any type of sensor that measures a vector quantity.
Examples of such sensors are: accelerometers, gyroscopes, magnetometers, gravimeters etc.
In many cases, the offsets on the output of these sensors pose a serious problem. Especially when the outputs of the sensors are numerically integrated, unknown offsets may be disastrous.
A patent application for this new DriftLess⁺ technology is being prepared.

The basic operation of the technology is depicted in the figure on the right. Two sensor units are used. Each sensor unit consists of one or more sensing axis. In general, each sensor unit has three sensing axes (x, y and z).
Both sensor units are rotated by some rotation mechanism, controlled by a processing unit. Both sensor units measure the same input quantity x̅ (the line above the x denoting its a vector quantity).
The sensor output signals s̅₁ and s̅₂ are processed by the processing unit. This processing unit estimates the offsets and compensates for it, resulting in the corrected sensor output y̅.
More detailed information about the basic operation of the DriftLess technology can be found in a paper. Another paper describes the benefits of this technology in an INS/GPS integrated navigation system.

Using this technology, the offsets of the sensors are estimated. This is done continuously during operational use. If the offsets change, for instance caused by sensor temperature changes, the changing offset is immediately estimated and compensated for. The results is that the offsets are virtually completely eliminated, leaving only a small remainder.

Images on the left: prototype of a DriftLess⁺ corrected sensor system. The unit outputs acceleration and rotation rate in 3 axes (x,y &z). In this specific version of the prototype, only the biases of the x & y axes are corrected. So, this version of the DriftLess⁺ system is a 2D-version, which is particularly usefull in near 2D-navigation systems found in most AGV's. The technology itself obviously allows the estimation of the biases of all three axes.
This prototype is based on very low-budget sensor chips (MPU6000), but the resulting unit acts as an IMU with very low residual bias (gyro's less then 10 deg/hr and accelerometers less then 1 mg). Dimensions of the unit are approximately 7x5x4 cm.
The technology clearly shows how a high performance IMU can be build, based on low-performance (and low-budget) sensor chips.

Images below: Allan Variance (AV) of the uncorrected and bias corrected gyroscopes. The Allan Variance is a measure of the stability of the biases. Initially, the method was developed to characterise the stability of crystal oscillators, but nowadays it is also widely adopted to characterise the stability of inertial sensor biases. As can been seen, the uncorrected gyros have have increasing bias variations at long averaging times. After compensation, the slow/long term bias variations including static bias is eliminated.


The DriftLess⁺ technology has been tested in an AGV-application. A small demonstrator-AGV has been made with a navigation system that only used odometry (wheel sensors). As expected, the navigation solution rapidly drifted away, in the order of 10 cm/m, or 10% Distance Travelled (DT). Using DriftLess⁺ corrected gyros, the navigation solution drasticly improved to an order of 0.2% DT. Results of the demonstration can be found in a white paper.