---

For the so-called dead-reckoning navigation of an autonomous mobile robot two measurements systems were available. The first system used incremental encoders mounted on the two driven wheels of the robot to measure its position by odometers, while the second system made use of the measurements from inertial sensors. Each measurement system has its own advantages and disadvantages with regard to accuracy. This project targeted the  optimization of the combined performance of these two systems by:

1. Integration of two different fiber optic gyroscopes in the hardware and software of the robot (VME based real-time computer system programmed in C). Comparison of the performance of the two sensors referring to a third, more accurate, dynamically tuned gyroscope, already integrated in the robot.
2. Development of the signal electronics for a navigational grad accelerometer and engineering of a test rig to collect experience with this sensor. Basic experimentation with the accelerometer on the test rig as well as on the robot itself.
3. Building up of two mathematical models for the study of the system dynamics and the sources of errors in the navigation systems. The first model is a third order linear one and considers no slip between the two driven wheels and the ground, while the second model is a fifth order nonlinear model that considers slip between the two driven wheels and the ground.
4. The parameters of both models were determined through a series of identification experiments. In order to obtain the complete nonlinear model parameters it was necessary to give special attention to the relationship between the slip velocity of a driven wheel and the resulting ground force at this wheel contact point with the ground.
5. The influences of the different error sources in the two navigation systems were studied through intensive simulation experiments. To be able to carry out these simulation experiments it was necessary to integrate the different error mechanisms in both navigational systems in the dynamic models. This integration has led to an expansion of the system basic model, in order to consider all known error sources. Then it was possible to answer the questions, which error sources have the greater effects on the accuracy of the navigation and which of these can be compensated for.
6. Based on the results of the above work a dynamic model for the correction of the effects of the slip on the navigation accuracy was integrated in the navigation software of the robot and it was then successfully proofed.
7. Considering the whole experience collected during the work a suggestion for a calibration procedure in the form of an „Alignment Procedure“ of the complete navigation system was suggested.
8. A simulation comparison between the accuracy of both navigation systems was carried out.

For a dedicated presentation on the subject please contact us.

The expression "Dead Reckoning" refers to navigation systems where the position of the moving vehicle is calculated through the timely integration of error affected measurements (such as rotational rates and linear accelerations) with no direct position measurement or reference point. This type of navigation is common for ships, as well as airplanes and missiles. An important feature of this type of navigation is the continuous accumulation of the error with time.