List of Symbols
INPUT LAYER HA(X4)
4.6 Demonstrations of the MANFIS Path Controller
4.6.3 Comparison of the Design MANFIS Navigational Controller with other Models
In this part, a comparative study has been carried out from the developed MANFIS approach with results proposed from the other techniques. The simulation study has been performed to show the performance of the proposed hybrid system. The size dimension of simulation platforms are considered as no. of units and each unit is in millimeter (mm). In the first comparative study, the proposed MANFIS navigational controller has been implemented in an environment shown in Figure 4.25(a). It has been noticed that using the developed method the robot covers a shorter path compared to the result proposed by Shi et al.[116]. However, it can be seen from the Figure 4.25(a) that the fuzzy-neural method proposed by author produces a smoother path to reach the goal, but it doesn’t guarantee a shorter compared to the MANFIS technique results shown in Figure 4.25(b).
In the second comparative study, the developed hybrid path controller has been applied in a similar environment as stated by the Mo et al.[87]. They used the fuzzy behaviour based navigational controller to navigate the robot safely in very small gaps as illustrated in Figure 4.26(a). From the simulation result, it is clearly seen that where the robot is making turns, there are some instances where the mobile robot closer to an obstacle than the required safe distance. This problem has been taken care of by the
proposed hybrid algorithm and depicted in the Figure 4.26(b). The effectiveness of both comparative studies has been measured on the basis of path length and tabulated in Table- 4.11. A comparative study has been carried out between ANFIS and MANFIS navigation systems in terms of path lengths to demonstrate the effectiveness of the path planner and tabulated in Table-4.12.
Figure 4.25 (a) Navigation path framed for a single mobile robot to reach target by Shi et al. [116].
Figure 4.25 (b) Navigation path framed for a single mobile robot to reach target using developed MANFIS method.
Figure 4.26 (a) Navigation path framed for a single mobile robot to reach target by Mo et al. [87].
Figure 4.26(b) Navigation path framed for a single mobile robot to reach goal using developed MANFIS method.
Table 4.11 Comparison of results in terms of path length.
Sl
No. Environment types
Path length from current system (in ‘cm’) Path length of reference model (in‘cm’) % of deviation 1 Complex environment with long obstacles Figures 4.25(a) and 4.25(b)
7.6 8.8 13.64
2 Maze environment Figures 4.26(a) and 4.26(b)
7.9 9.3 15.05
Table 4.12 Comparison of ANFIS and MANFIS results in terms of path length.
Sl
No. Environment types
Path length using ANFIS system (in ‘cm’) Path length using MANFIS model (in‘cm’) %of deviation 1 Scenario-1 6.34 6.61 4.08 2 Scenario-2 8.30 8.59 3.37 3 Scenario-3 7.78 8.10 5.18 4 Scenario-4 10.25 10.49 2.28
4.7 Summary
This chapter has described the implementation of ANFIS and MANFIS techniques for the mobile robot navigation.
The following salient features are drawn based on the simulation and experimental results using ANFIS technique.
The developed novel ANFIS navigational controller has been successfully used to control the robot in a highly cluttered environment.
With the help of this proposed navigation system, the robots are able to perceive the environment condition and reach the goal successfully.
In the simulation results, it clearly observed that various reactive behaviours such as obstacle avoidance, barrier following and target seeking have been performed by the proposed navigational controller.
A series of practical tests have been carried out with a real developed robot to show the efficacy and effectiveness of the proposed navigation algorithm. They are found to be in good agreement. The percentages of errors are found to be within 7% for both path length and time taken for the robot to reach the target and tabulated in Tables 4.4 and 4.5 respectively.
A comparative study is carried out between the performance of the proposed navigation and those of obtained by authors [87, 116] in simulation mode. It has been observed that the proposed navigation method provides better results compared to other techniques. The performance of the comparison study is mainly measured in terms of path length and tabulated in Table 4.6.
The following conclusions are drawn on the basis of simulation and experimental results using MANFIS methodology.
The proposed methodology has been successfully implemented for solving the navigational problem of a mobile robot in an unknown or partially known static environment.
The efficacy of the proposed navigation technique has been demonstrated through various exercises. During experimentation, it has been observed that the proposed path planner has the ability to avoid obstacles in a cluttered environment.
The real time experimental tests have been performed to validate the developed navigation system. The percentages of errors are found to be within 7% for both path
length and time taken for the robot to reach the target and tabulated in Tables 4.9 and 4.10 respectively.
A comparative graphical study has been demonstrated to find out the effectiveness of the proposed navigational controller and percentage of deviation is shown in the Table-4.11.
Finally, it has been inferred that the proposed path planning system produce closer results to the ANFIS navigation system.
In the consequent chapters Cuckoo search algorithm and Invasive weed optimization algorithm have been investigated and examined as standalone methods. These algorithms are then combined and hybridized with ANFIS to produce better navigational controller for mobile robots.