One of the purposes of the Fast Traverse Rover project is to test scientific payloads in the central compartment. The first payload intended for the platform is an autonomous shear vane instrument. A shear vane is a manual instrument that is pressed into soil and twisted to determine the shear load the soil can take before giving way. A rover equipped with the ability to analyze if the terrain can withstand the weight of driving is capable of making extremely well informed decisions in the face of uncertainty.
The objective of this project was to construct a proof of concept device while remaining as inexpensive as possible. Therefore, most of the mechanical and electrical hardware was already on hand.
The main functionalities of the payload are to press the shear vane into the soil with a given load and twist the instrument until it breaks free of the soil. A curve of the soil's Mohr-Coulomb behavior can be generated by applying a range of normal loads applied to the device and compiling the data points. Beam load cells were used to read the normal and torsional forces applied by the apparatus to the instrument. The instrument dial was digitized with a potentiometer to confirm that the torsional loads were accurate. While the normal force was applied by the linear servo and torsional force by a stepper motor.
The MATLAB software I wrote to coordinate all of these devices and collect the data worked. Here are the resulting readings from the control samples and field tests. More information can be found on the paper itself. (Paper PDF)
However, through this experience I learned that there are several improvements I would make for a rover mounted prototype.
First of all I would select equipment with a higher baud rate to reduce system lag, or utilize a different programming language entirely to avoid the latency experienced during experiments. However, this may be alleviated with more tuning of the system. Second, I would aim to acquire and integrate a single load cell capable of normal and torsional readings in order to simplify and further reduce the size of the system. Third, I would integrate more compliance to the system so that the device can remain level to the ground. Experiments should that is was sensitive to operating at an angle and performed poorly on sloped surfaces. Further testing would definitely be needed to tune the system for performance on sand in addition to the clay & silt soils that it already performed well on.
All in all a relatively successful proof of concept, but much more work to be done if we are to deploy it and derive more conclusive results.