International Journal of Adaptive and Innovative Systems (5 papers in press)

Regular Issues

• Adoption Dynamics of the System of Rice Intensification by smallholder farmers in Tanzania: A Case of Mkindo Irrigation Scheme  
  by Asheri Mwidege 
  Abstract: Intermittent irrigation is among the activities performed to minimise water usage when adopting the system of rice intensification (SRI). Little information is known about the drivers toward the adoption or dis-adoption of SRI practices by subsistence farmers at the Mkindo Rice Irrigation Scheme in Tanzania. An experimental research design and expert sampling plan were employed in which cross-sectional data of 100 SRI and non-SRI participants. Descriptive statistics information was obtained using the SPSS package. Survey results showed high grain yield (58.93%) and increased return to labour (51.79%) as key factors that influenced rice farmers to adopt SRI. Contrary, lack of training (79.54%), awareness (77.27%) and skills acquisition (52.27%) on SRI practices were factors that negatively influenced the adoption of SRI practices by non-adopters. Furthermore, SRI practices in masika, accessible extension services had a statistically significant positive effect on farmers’ income contrary to vuli practices and famer gate prices at p < 0.01 level. It was concluded that high grain yield was evidenced in masika while accessing extension services thus increased return to labour influenced smallholder rice farmers to adopt SRI practice.
  Keywords: adoption; dis-adoption; systems of rice intensification; economic analysis; Mkindo irrigation scheme; Tanzania; system of rice intensification; SRI.
  DOI: 10.1504/IJAIS.2022.10050551
   
  
• Self-Driving Cars Perception: Transforming the Morphology of the Automotive Industry  
  by Ihssane Bouasria, Walid Jebrane, Nabil Elakchioui 
  Abstract: Several emerging computing algorithms and methods including deep learning, deep reinforcement learning, and terms like end-to-end learning, besides their applications in the computer vision, control, robotics, or automotive fields, on one hand, seem incomprehensible, threatening, and hard to be explained or contained in one form of architecture, and on the other hand, the public tends to fear the things they do not understand and approach these topics with precaution and skepticism, leading to a significant gap between theoretical and empirical research goals. Meanwhile, self-driving cars are increasingly becoming fully automated and manufacturers are racing to enter the mainstream. This article examines the key concept of self-driving cars. It begins by highlighting the importance of autonomous cars in reshaping the infrastructure. Major methods and techniques found in various sets of literature are identified and detailed. Existing research concerning the perception system and object detection is then outlined. Finally, remarks and suggestions are presented.
  Keywords: autonomous vehicles; computer vision; perception; deep learning; object detection.
  DOI: 10.1504/IJAIS.2022.10051435
   
  
• Review of Effective Machine learning Methods for disease prediction  
  by Aicha Oussous, Abderrahmane Ez-zahout, Soumia Ziti, Oussous Ahmed 
  Abstract: There is a wealth of data available nowadays, including health data. Artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL) are required to intelligently evaluate this data and generate the corresponding creative applications. In actuality, these techniques may identify models from a range of medical data sources, predict diseases, and help with rapid decisions that enhance patient safety and treatment quality. There is a wealth of data available nowadays, including health data. Artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL) are required to evaluate this data and generate the corresponding creative applications intelligently. In actuality, these techniques may identify models from various medical data sources, predict diseases, and help with rapid decisions that enhance patient safety and treatment quality. This review intends to demonstrate the potency of several machine learning approaches in identifying diseases. This study looked at the accuracy of several machine learning techniques, including single, hybrid and ensemble hybrid algorithms. The results indicate that the hybrid ensemble machine learning technique is the best when comparing the accuracy of each method.
  Keywords: machine learning; hybrid model; hybrid ensemble model; disease prediction.
  DOI: 10.1504/IJAIS.2022.10051514
   
  
• Fast Terminal Sliding Mode Lateral Control Combined with Artificial Neural Networks for Autonomous Automobile  
  by Najlae Jennan, El Mehdi Mellouli 
  Abstract: This work presents the fast terminal sliding mode control method combined with neural networks technique that concerns the modelling and the controlling of a nonlinear system. First of all, we propose the neural network method for the aim to approximate the unmodelled lateral dynamics of autonomous automobile. Then, we add an auxiliary term to the control law to reduce the disturbances and modelling errors. Finally, we use the particle swarm optimisation method to optimise the important coefficients in the control law. The Lyapunov approach is used to study the system stability. The efficiency of the proposed strategy is shown in the results simulation.
  Keywords: nonlinear system; fast terminal sliding mode control; FTSMC; neural networks; particle swarm optimisation; PSO; autonomous vehicle; lateral control.
  DOI: 10.1504/IJAIS.2022.10051584
   
  
• DRL keeps Autonomous Vehicles on Track: Efficient Motion Planning and control model  
  by Walid Jebrane, Ihssane Bouasria, Nabil Elakchioui 
  Abstract: Mobile robots have greatly contributed to the intelligent development of human society in recent decades. Similarly, the movement planning policy is the essential element for autonomous vehicles. This article reviews different approaches to movement planning, especially those that incorporate deep reinforcement learning (DRL) in the unstructured environment. First, we review the basics of DRL elements and their most notable developments. Next, we take an in-depth look at the most important research by reporting and discussing a detailed survey of recent technological advances in motion planning for single and multi-agent autonomous vehicles. Finally, we covered potential technology trends for AV motion planning and the major limitations that hinder its development.
  Keywords: autonomous vehicles; deep reinforcement learning; DRL; artificial intelligence; motion planning.
  DOI: 10.1504/IJAIS.2022.10051629