3.6 Advanced 3D Attitude Representation

Although the quaternion display yields a great deal of information about the rate of convergence and the size of attitude tracking errors, it is still very difficult to form a mental picture of the dynamic motion of the vehicle by simply observing the quaternion elements. Using the desired and estimated quaternions to specify the attitude of 3 dimensional graphic representations of the vehicle and allowing these pictures to move in real time produces an even deeper understanding of system performance. Directly observing the vehicle in the water allows the pilot to easily spot some basic problems such as limit cycling. However, it is more difficult to spot some of the more subtle effects such as a rotational buoyancy offset that causes the vehicle to always converge with the nose slightly down, or a dynamic interaction that causes the vehicle to roll slightly during a yaw step response. For effects such as these, it is helpful to have a display where the desired and estimated reference frames are superimposed on top of each other, so the motion of one with respect to the other may be directly observed.

Figure 3-6 Advanced 3D Attitude and Data Display

Using the World Modeler (a 3D world modeling toolkit created by the author), a 3D model of Ranger NBV was created. Using a video camera, images of the external surfaces of the vehicle were digitized, and then placed on the model in the proper locations. This texture mapped model was then used to represent the estimated attitude of the vehicle. Although at first glance texture mapping the surface of the model seem to have no purpose other than to make the display look pleasing, experience has shown that it is easier to perceive subtle motions of the vehicle when using the texture mapped display. The desired vehicle attitude is represented by a wireframe model of the same shape.
Often during a pool test, it is desirable to view the vehicle from a specific angle to gain a greater understanding of the vehicle performance. While physical camera views are limited, the 3D display allows the pilot to move the virtual camera to the location that is best suited for observing a particular maneuver. The mouse is used to place the camera with respect to the vehicle.
The 3D display window is also capable of displaying additional information in the form of strip charts. The attitude error (the Euler angle from (4.6)) is displayed, and may be scaled both in time, and in magnitude. It may also be filtered in real time to smooth the overall display which filters out the display of high frequency effects and yields insight into overall convergence trends.
Strip charts of the desired and actual vehicle torques are also displayed. This makes it easy to determine when the controller is asking the thrusters to generate more thrust than they are capable of producing. This condition, called thruster saturation, can severely degrade vehicle performance. It is helpful to identify in real time when saturation is occurring so that the relationship between performance and saturation can be observed, and appropriate actions can be taken to reduce its effects.
Similarly, strip charts of the three elements of the angular velocity error vector are also available. This data also may be scaled along the magnitude and time axes. This data is useful for observing phase lag, and sensor bias.