Tercera generación: nacimientos entre 1980 y 1990

3.3

Single Locomotor Test Beds

As seen in the literature review from Section 2.3, it is common practice in Terramechanics research and a useful tool for the analysis and semi-empirical modelling of robot-soil inter- action to use simplified test beds with a single locomotor [104]. Such test beds facilitate repeatable control of testing variables in preparation for final validation experiments on fully mobile rover prototypes.

3.3.1 Single Wheel-Leg Test Bed

A Single Wheel Test Bed (SWTB) used in-house for previous research projects [152] was adapted to carry a driven wheel-leg with the early WLSIO sensor system prototype described in Sub-section 3.2. An extruded aluminium frame supports two precision shafts and a platform is linked to them via two linear PTFE bearings. This platform acts as a moving carriage, translated along the shafts by a motorised timing belt. Additional precision shafts are linked vertically to the moving carriage through PTFE bearings to allow freely sinking motion. At the lower end of the vertical precision shafts a rigid assembly is attached containing the driven wheel-leg, the WLSIO prototype and the electronics to control the speed of both DC motors driving the wheel-leg and the moving carriage. The wheel-leg assembly is driven along a 1 m long, 75 cm wide and 20 cm deep bed of regolith as it sinks into it.

Different levels of slip can be simulated by controlling the relative speeds of the timing belt and the wheel-leg. The angular position and speed of the wheel-leg is measured by the previously described absolute angular position encoder. The same type of encoder is mounted on the sprocket wheel driving the timing belt to measure the linear displacement and speed of the moving carriage. The set-up with labels on all the components is shown in Fig. 3.5.

The SWTB set-up is useful for initial experimentation, but is not very representative of the tilting body frame motion observed in hybrid wheel-legs due to their rimless, irregular shape. To better simulate the motion of the wheel-leg during testing the linear DoFs are replaced, resulting in the Single Wheel-Leg Test Bed (SWLTB) pictured in Fig. 3.6.

3.3. Single Locomotor Test Beds

(a) (b)

Figure 3.5: Diagram with degrees of freedom (a) and labelled image (b) of the SWTB

Instead of the timing belt, a driving wheel is used to pull the moving carriage, which rolls over two guiding rails that extend over a 5 m long, 1 m wide and 20 cm deep volume of re- golith. The vertical linear bearings are substituted by a single passive rotary DoF, mounted on the moving carriage with its rotation axis horizontally orthogonal to the direction of driving. With the new set-up, the reference frame of the wheel-leg assembly tilts as the

(a) (b)

3.3. Single Locomotor Test Beds

wheel-leg rolls, rather than just translating up and down in the vertical direction. Testing different slip conditions is still possible by regulating the relative speeds of the wheel-leg and the carriage driving wheel.

Due to the lack of a timing belt or encoder on the carriage-driving wheel, the absolute linear position and speed of the carriage is measured by tracking the known position of fixed markers from cameras mounted on the moving carriage. The same SBC and camera module used for the WLSIO BC are used for this purpose, in combination with the ARUCO library [153].

Before each test is performed both using the SWTB and SWLTB set-ups in the laboratory the soil was prepared using the consistent raking method described in Sub-section 3.5.4, so as to re-homogenise the structure of the soil disturbed by the previous test and minimise the variability of soil properties between tests due to different compaction states. Multiple repetitions of each set of testing conditions were carried out, permuting the order of the tests to further reduce any potential influence of successive soil re-settings. In addition, a vibration device was dragged along the soil surface for a sub-set of the testing conditions to achieve higher compaction levels and test the ability of the WLSIO system to differentiate between medium and dense configurations of the same soil type.

To evaluate the consistency of the soil preparation method used DCP tests were carried out every 0.5 m along the SWLTB. The obtained penetration per impact vs. total penetration curves were then analysed. If the values were outside the expected variability seen in small scale calibration DCP tests, the soil was re-prepared before carrying out the wheel-leg forward driving test.

3.3.2 Field Testing Set-up and Environment

Testing in more realistic conditions is important for an appropriate evaluation of the per- formance of the sensor system and to identify sources of errors that do not occur in the con- trolled laboratory environment. Field tests provide natural lighting conditions and terrain profiles, with heterogeneous soil compositions, slopes and irregular surfaces. The SWLTB set-up is designed to be easily adapted for attachment with any mobile platform.

3.3. Single Locomotor Test Beds

(a) (b)

Figure 3.7: SWLTB attachment mounted on the SMART rover for field testing, with both wheel-leg configurations used: LTF (a) with CIF and (b) with FRF

Using the in-house wheeled micro-rover SMART [154], experiments were carried out in the West Wittering beach in the Southern English coast. The passive rotary DoF of the SWLTB was linked with the chassis of the SMART rover as seen in Fig. 3.7. Two different wheel-leg set-ups were used, combining LTF with CIF and FRF shown in Fig. 3.7 (a) and (b) for CIF and FRF respectively. The wheel-leg speed command and WLSIO data logging interfaces were easily set with the OBC of the SMART rover, since it operates using the same SW framework as the WLSIO system.

Figure 3.8: Diagram of the distribution of different types of terrain in the field testing site and the paths of the tests carried out (left) and images of the different types of terrain (right)