A vision based application is proposed for a line following mobile robot. A low-cost Webcam is used as the sensor and the image buffers are processed via a customized image segregation method to output necessary information for the mobile robot's controller under uncontrollable lighting condition. The task is to allow the mobile robot to navigate through a predefined path marked by a white line on a dark green floor surface. Experimental results shows that the mobile robot successfully able to navigate throughout the provided path. This technique is robust, reliable and easily modified for used in any other applications, such in industrial applications where costly sensors and transducers are normally used.

The increasing severity of the labour shortage problem in the Malaysian palm oil industry has created a need to explore other avenues for harvesting oil palm fresh fruit bunches (FFBs) such as through autonomous robots’ deployment. However, the first step in using an autonomous system to harvest FFBs is to identify which FFBs have become ripe and are ready to be harvested. In this work, we reviewed previous and current methods of identifying the maturity of fresh fruit bunches as found in the literature. The different methods were then compared in terms of the types of sample data used, sensor modalities, and types of classifiers used with a particular focus on the feasibility of each method for on-field application. From the 51 papers reviewed, which include a total of 11 unique approaches, it was found that the most feasible method for detecting ripe FFBs in the field is a combination of computer vision and deep learning. This system has the advantages of being a noncontact approach that is low cost while also being able to operate in real time with high accuracy.

No abstract

The objective detection of muscle fatigue reports the moment at which a muscle fails to sustain the required force. Such a detection prevents any further injury to the muscle following fatigue. However, the objective detection of muscle fatigue still requires further investigation. This paper presents an algorithm that employs a new fatigue index for the objective detection of muscle fatigue using a double-step binary classifier. The proposed algorithm involves analyzing the acquired sEMG signals in both the time and frequency domains in a double-step investigation. The first step involves calculating the value of the integrated EMG (IEMG) to determine the continuous contraction of the muscle being investigated. It was found that the IEMG value continued to increase with prolonged muscle contraction and progressive fatigue. The second step involves differentiating between the high-frequency components (HFC) and low-frequency components (LFC) of the EMG, and calculating the fatigue index. Basically, the segmented EMG signal was filtered by two band-pass filters separately to produce two sub-signals, namely, a high-frequency sub-signal (HFSS) and a low-frequency sub-signal (LFSS). Then, the instantaneous mean amplitude (IMA) was calculated for the two sub-signals. The proposed algorithm indicates that the IMA of the HFSS tends to decrease during muscle fatigue, while the IMA of the LFSS tends to increase. The fatigue index represents the difference between the IMA values of the LFSS and HFSS, respectively. Muscle fatigue was found to be present and was objectively detected when the value of the proposed fatigue index was equal to or greater than zero. The proposed algorithm was tested on 75 EMG signals that were extracted from 75 middle deltoid muscles. The results show that the proposed algorithm had an accuracy of 94.66% in distinguishing between conditions of muscle fatigue and non-fatigue.

Fresh Fruit Bunch (FFB) is the main ingredient in palm oil production. Harvesting FFB from oil palm trees at its peak ripeness stage is crucial to maximise the oil extraction rate (OER) and quality. In current harvesting practices, misclassification of FFB ripeness can occur due to human error, resulting in OER loss. Therefore, a vision-based ripe FFB detection system is proposed as the first step in a robotic FFB harvesting system. In this work, live camera input is fed into a Convolutional Neural Network (CNN) model known as YOLOv4 to detect the presence of ripe FFBs on the oil palm trees in real time. Once a ripe FFB is detected on the tree, a signal is transmitted via ROS to the robotic harvesting mechanism. To train the YOLOv4 model, a large number of ripe FFB images were collected using an Intel Realsense Camera D435 with the resolution of 1920×1080. During data acquisition, a subject matter expert assisted in classifying the FFBs in terms of ripe or unripe. During the testing phase, the result of the mean Average Precision (mAP) and recall are 87.9 % and 82 % as the detection is fulfilled the Intersect over Union (IoU) with more than 0.5 after 2000 iterations and the system operated at the real-time speed of roughly 21 Frame Per Second (FPS).