Control Nonholonomic Mobile Robot with Hybrid Sliding Mode/Neuro Fuzzy Controller
Many works has been done in the mobile robots research domain, resulting different methods to enhance the performance of the mobile robot. This article will adopt a hybrid approach to improve the performance of a path tracking controller by designing an a
- PDF / 474,372 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 32 Downloads / 228 Views
Abstract. Many works has been done in the mobile robots research domain, resulting different methods to enhance the performance of the mobile robot. This article will adopt a hybrid approach to improve the performance of a path tracking controller by designing an algorithm that uses two methods: the first one is Sliding Mode [SM], which will have one of its parameters controlled by the second one based on neurofuzzy [NF]. Keywords: Sliding mode
Neural network Fuzzy logic Mobile robot
1 Introduction Various research laboratories have studied the mobile robot control domain, trying to optimize their performance [1] by seeking more effective new control methods [2, 3]. Various methods based on artificial intelligence [4, 5], genetic algorithms, mathematic or heuristics were used [6–8]. But while those methods were being experimented, a new hybrid approach was developed [9] in order to take the best of each methods. This is done by smartly combining them into one controller. During this work, we tried to optimize the performance of a path tracking controller for a non-holonomic mobile robot. The main controller is based on a robust sliding mode controller [10] which we will try to optimize by dynamically change one of its variables instead of using fixed value. To achieve this we will use a neuro-fuzzy mode controller [11], this to ensures that the variable will have the optimum value.
2 System Presentation Considering that our system is a non-holonomic mobile robot, its kinematic model should be under the form (Fig. 1): © Springer International Publishing Switzerland 2016 Y. Tan et al. (Eds.): ICSI 2016, Part II, LNCS 9713, pp. 451–458, 2016. DOI: 10.1007/978-3-319-41009-8_49
452
M.N. Houam et al.
8 x_ ¼ v1 cosðhÞ > > > > > < y_ ¼ v1 sinðhÞ tanðuÞ > h_ ¼ v1 > > > lw > : u_ ¼ v2
ð1Þ
Fig. 1. Mobile robot projection in a 2D space
In order to simplify, the chained form of the system has been used. It is written as follows [12, 14]: 8 z_ 1 ¼ u1 > > > > > z_ 2 ¼ u1 z3 > > > > > < z_ 3 ¼ u1 z4 > ... > > > > > > z_ n1 ¼ u1 zn > > > : z_ n ¼ u2
ð2Þ
Having a kinematic model of 4th order, our chained form system is equivalent to: 8 z_ 1 > > > < z_ 2 > z_ 3 > > : z_ 4
¼ u1 ¼ u1 z 3 ¼ u1 z 4
ð3Þ
¼ u2
With 8 z1 > > > > < z2 > z3 > > > : z4
¼x ¼y ¼ tanðhÞ ¼ tanðuÞ=ðl cosðhÞ3 Þ
ð4Þ
Control Nonholonomic Mobile Robot
453
And 8 > < v1 ¼ u1 cosðhÞ 3 sinðhÞ > sin2 ðuÞ u1 þ l cos3 ðhÞ cos2 ðuÞ u2 : v2 ¼ l cos2 ðhÞ
ð5Þ
3 Sliding Mode/Neurofuzzy Controller Based on the sliding mode method [2, 13], a sliding mode controller parameter has been optimized by inserting a neuro-fuzzy controller which will compute dynamically its value. So the main controller (Sliding Mode, SM) will regulate the path tracking, and the second controller (Neuro-fuzzy, NF) will optimize the chosen parameter. As follow, the used control method (SM controller) [10] and the hybrid controller are presented. 3.1
Sliding Mode Controller
• Path tracking by sliding mode: For the path tracking using sliding mode,
Data Loading...
