The use of Internet of things (IoT)-based physical sensors to perceive the environment is a prevalent and global approach. However, one major problem is the reliability of physical sensors’ nodes, which creates difficulty in a real-time system to identify whether the physical sensor is transmitting correct values or malfunctioning due to external disturbances affecting the system, such as noise. In this paper, the use of Long Short-Term Memory (LSTM)-based neural networks is proposed as an alternate approach to address this problem. The proposed solution is tested for a smart irrigation system, where a physical sensor is replaced by a neural sensor. The Smart Irrigation System (SIS) contains several physical sensors, which transmit temperature, humidity, and soil moisture data to calculate the transpiration in a particular field. The real-world values are taken from an agriculture field, located in a field of lemons near the Ghadap Sindh province of Pakistan. The LM35 sensor is used for temperature, DHT-22 for humidity, and we designed a customized sensor in our lab for the acquisition of moisture values. The results of the experiment show that the proposed deep learning-based neural sensor predicts the real-time values with high accuracy, especially the temperature values. The humidity and moisture values are also in an acceptable range. Our results highlight the possibility of using a neural network, referred to as a neural sensor here, to complement the functioning of a physical sensor deployed in an agriculture field in order to make smart irrigation systems more reliable.

Over recent years, the demand for supplies of freshwater is escalating with the increasing food demand of a fast-growing population. The agriculture sector of Pakistan contributes to 26% of its GDP and employs 43% of the entire labor force. However, the currently used traditional farming methods such as flood irrigation and rotating water allocation system (Warabandi) results in excess and untimely water usage, as well as low crop yield. Internet of things (IoT) solutions based on real-time farm sensor data and intelligent decision support systems have led to many smart farming solutions, thus improving water utilization. The objective of this study was to compare and optimize water usage in a 2-acre lemon farm test site in Gadap, Karachi, for a 9-month duration, by deploying an indigenously developed IoT device and an agriculture-based decision support system (DSS). The sensor data are wirelessly collected over the cloud and a mobile application, as well as a web-based information visualization, and a DSS system makes irrigation recommendations. The DSS system is based on weather data (temperature and humidity), real time in situ sensor data from the IoT device deployed in the farm, and crop data (Kc and crop type). These data are supplied to the Penman–Monteith and crop coefficient model to make recommendations for irrigation schedules in the test site. The results show impressive water savings (~50%) combined with increased yield (35%) when compared with water usage and crop yields in a neighboring 2-acre lemon farm where traditional irrigation scheduling was employed and where harsh conditions sometimes resulted in temperatures in excess of 50 °C.

Weed management is becoming increasingly important for sustainable crop production. Weeds cause an average yield loss of 11.5% billion in Pakistan, which is more than PKR 65 billion per year. A real-time laser weeding robot can increase the crop’s yield by efficiently removing weeds. Therefore, it helps decrease the environmental risks associated with traditional weed management approaches. However, to work efficiently and accurately, the weeding robot must have a robust weed detection mechanism to avoid physical damage to the targeted crops. This work focuses on developing a lightweight weed detection mechanism to assist laser weeding robots. The weed images were collected from six different agriculture farms in Pakistan. The dataset consisted of 9000 images of three crops: okra, bitter gourd, sponge gourd, and four weed species (horseweed, herb paris, grasses, and small weeds). We chose a single-shot object detection model, YOLO5. The selected model achieved a mAP of 0.88@IOU 0.5, indicating that the model predicted a large number of true positive (TP) with much less prediction of false positive (FP) and false negative (FN). While SSD-ResNet50 achieved a mAP of 0.53@IOU 0.5, the model predicted fewer TP with significant outcomes as FP or FN. The superior performance of the YOLOv5 model made it suitable for detecting and classifying weeds and crops within fields. Furthermore, the model was ported to an Nvidia Xavier AGX standalone device to make it a high-performance and low-power computation detection system. The model achieved an FPS rate of 27. Therefore, it is highly compatible with the laser weeding robot, which takes approximately 22.04 h at a velocity of 0.25 feet per second to remove weeds from a one-acre plot.

The management of the abundant amount of used plastic bottle waste is a major concern nowadays, because it is a major contributor to landfills and overburdens waste processing facilities. Once disposed of, plastic can take centuries to break down, hence, recycling not only manages the waste efficiently, but it reduces the environmental impact and creates economic opportunities, as well. An incentive-based Reverse Vending machine (RVM) is an effective way to involve the general public in the management of plastic waste. The existing solutions are either very expensive, from a computation and cost perspective, or they lack the robustness and durability necessary for deployment. In this research, we have developed an incentive-based low-cost RVM for the collection, identification, classification, and sorting of used plastic bottles with the addition of a reward-based user application. The developed RVM includes a low-cost computing device running a lightweight image processing algorithm, sensors, camera, and a self-designed mechanical arm. To support the low computing device in the RVM, a lightweight MobileNet model has been trained using transfer learning. A dataset of 10,983 pet bottle images was collected using a camera installed inside the machine for the classification model. The results of the study show that MobileNet achieved 99.2% testing accuracy, which is better than the existing bottle-classification approaches. Moreover, the weight of the developed model was only 12 Mb, which is fourteen and six times less than inceptionV3 and Res-Net (Residual Neural Networks), respectively. Furthermore, the developed RVM costs a fraction of the price, compared to the existing solutions. Two versions of the machine have been deployed at a University for more than 6 months, collecting over 650 kg of plastic waste.

Weed management has become a highly labor-intensive activity, which is the reason for decreased yields and high costs. Moreover, the lack of skilled labor and weed-resistant herbicides severely impact the agriculture sector and food production, hence increasing the need for automation in agriculture. The use of agricultural robots will help in the assurance of higher yields and proactive control of the crops. This study proposes a laser-based weeding vehicle with a unique mechanical body that is adjustable relative to the field structure, called the Robot Operating System (ROS) based robust control system, and is customizable, cost-effective and easily replicable. Hence, an autonomous-mobile-agricultural robot with a 20 watt laser has been developed for the precise removal of weed plants. The assembled robot’s testing was conducted in the agro living lab. The field trials have demonstrated that the robot takes approximately 23.7 h at the linear velocity of 0.07 m/s for the weeding of one acre plot. It includes 5 s of laser to kill one weed plant. Comparatively, the primitive weeding technique is highly labor intensive and takes several days to complete an acre plot area. The data presented herein reflects that implementing this technology could become an excellent approach to removing unwanted plants from agricultural fields. This solution is relatively cost-efficient and provides an alternative to expensive human labor initiatives to deal with the increased labor wages.