Test 2

In this flight test, a circling flight under wind distance is carried out to demonstrate the capability of our proposed ESO based controller against the wind disturbance. Two 380 watts fans with diameter of 750 mm are set in the flying arena. A wind field is established where the black arrow represents wind speed and direction as depicted in Figs. 4.7(a) and 4.7(b). The maximum wind speed is up to 5m/s measured by a digital anemometer AS8556.

The quadrotor UAV travels three times along the desired circle for each scenario. The radius of the circle path is 0.8 nr and the flight speed is around 0.5 m/s. It can be seen from Fig. 4.7(a) that the actual trajectories are deviated to the northeast while the flight trajectories and the preset circle path overlap precisely in Fig. 4.7(b). The experimental performances demonstrate the effectiveness of the proposed ESO method against wind disturbance.

Test 3

To check whether the ESO only based controller can compensate the cable- suspended-payload disturbance for precise trajectory tracking, two circling flight tests with payload oscillating disturbance are conducted. Uj is approximately equal to 1/T where T is the period of a circling flight . To increase the payload oscillation amplitude during the UAV flight, the flight speed is increased to around 1.26 m/s and a, is approximately equal to 0.25 s^-1 and all of the other configurations are the same as Test 2.

As illustrated in Figs. 4.8(a) and 4.8(b), the actual flight trajectories in both cases evenly proportionally deviate from the preset circle trajectory due to the centrifugal force of the payload. No significant improvement has been achieved as long as only ESO is exploited. Since the UAV flies along a circle, the cable-suspended-payload disturbance can be modelled as a kind of disturbances with specific frequency. In such a situation, DO based controller should be involved.

Test 4

The results in three tests demonstrate the flight performances by using a single observer in different scenarios with only one disturbance. To demonstrate the capability of our proposed strategy against both payload oscillating disturbance and wind disturbance, four circling flight experiments under these two hybrid disturbances are carried out. The flight speeds in these tests are all set around 1.26m/s, <х_г- is approximately equal to 0.25s^_1 and the radius of the desired circle trajectory is 0.8 m as well. The quadrotor UAV flies three times along the desired path.

FIGURE 4.7: Circling flight under wind disturbance of Test 2: (a)

and (b) depict the flight trajectories without/with ESO in x and у directions, respectively.

FIGURE 4.8: Circling flight under payload disturbance of Test 3: (a)

and (b) depict the flight trajectories without/with ESO in x and у directions, respectively.

It can be observed from Fig. 4.9(a) and 4.9(b) that there exist significant trajectory deviations from the desired circle in these two scenarios. In comparison of Fig. 4.9(a), the shape of actual trajectories in Fig. 4.9(b) remains the same and bias is almost constant. This situation indicates that the ESO based controller has better wind resistance. Trajectories in Fig. 4.9(c) deviate less from the desired path when comparing to that in Fig. 4.9(b). It is depicted that the proposed DO based controller plays a significant role in attenuating the payload oscillating disturbance while the northward offset also shows that its wind resistance is not significant. Different from the above three cases, the

FIGURE 4.9: Circling flight under both payload oscillating and wind disturbances of Test 4: (a), (b), (c), and (d) depict the flight trajectories using classical method, ESO-based method, DO-based method and proposed MOBADC method, respective!)'.

flight trajectories in Fig. 4.9(d) have smaller trajectory deformation and offset, validating the effectiveness of our proposed strategy. The corresponding attitude information and motor control inputs by using MOBADC can be checked in Fig. 4.10. It can be seen from Fig. 4.10(a) that the attitude in response to three circling flights is relatively smooth. There exist some vibrations in roll angle when resisting wind disturbance and maintaining the desired trajectory. Under both payload oscillating disturbance and wind disturbance, the motors can be managed properly to handle these hybrid disturbances within an allowable range as depicted in Fig. 4.10(b). The flight tests results exemplify the applicability of the developed scheme in this study.

Assessment

To sum up, the statistical results (mean absolute error 7 = -f Х!Г=1 117' ^—

7d,i|| and standard deviation s = ^ _7ГЦ- Xa=i(ll7i - 7d,i|l ^- O')²) °f the preceding experiments are summarized in Table 4.1.

The quantitative indices are listed in Table 4.1. It is confirmed that the DO based controller improves the mean error and STD of the trajectory tracking by 61.32% and 64.96% in Test 1, which coincides with the merits of the proposed DO in attenuating cable-suspended-payload disturbance. The ESO based method narrows the mean error and STD by 64.47% and 55.38% in Test 2, which validates the effectiveness of the proposed ESO against wind disturbance. However, there is no significant improvement in Test 3, which shows that the ESO only based method is not able to sufficiently mitigate the payload oscillating disturbance. In Test 4, the mean error and STD by using MOBADC can be improved by 76.70% and 71.14%, respectively. These results outperform the classical common-used PD controller, ESO and DO only based controller, and demonstrate the effectiveness of the proposed strategy.

TABLE 4.1: Mean absolute error and standard deviation. (Unit: m)

FIGURE 4.10: Attitude information and motor control inputs of Test 4 by using MOBADC.

Conclusions

This chapter proposed a multiple observers based anti-disturbance control for quadrotor UAV against both cable-suspended-payload disturbance and wind disturbance. A DO based controller was proposed in the translational control loop which aims to attenuate the payload oscillating disturbance. With respect to the wind disturbance, two ESOs were designed in the translational and rotational control loop respectively to mitigate its effect on the flight trajectory tracking. In order to further enhance the anti-disturbance performance for multiple disturbances, the anti-disturbance control scheme was proposed in position loop by combining the preceding two observers. Extensive flight experiments validated the effectiveness of our proposed system.

Notes

From the literature, two types of anti-disturbance control schemes can be potentially applied to quadrotor UAVs. One is to design a robust controller for the quadrotor UAV, making the closed-loop system insensitive to possible disturbances. The other is to estimate the disturbance, and subsequently attenuate or even eliminate disturbances by appropriately utilizing disturbance information. In this chapter, we adopted the second method. And both the simulation results and flight experiments showed the effectiveness of proposed strategy. In the future, safety flight control of quadrotor UAV in terms of centre-of-gravity shift together with motors’ faults case can be investigated.