A fundamental challenge in autonomous vehicles is adjusting the steering angle at different road conditions. Recent state-of-the-art solutions addressing this challenge include deep learning techniques as they provide end-to-end solution to predict steering angles directly from the raw input images with higher accuracy. Most of these works ignore the temporal dependencies between the image frames. In this paper, we tackle the problem of utilizing multiple sets of images shared between two autonomous vehicles to improve the accuracy of controlling the steering angle by considering the temporal dependencies between the image frames. This problem has not been studied in the literature widely. We present and study a new deep architecture to predict the steering angle automatically by using Long-Short-Term-Memory (LSTM) in our deep architecture. Our deep architecture is an end-to-end network that utilizes CNN, LSTM and fully connected (FC) layers and it uses both present and futures images (shared by a vehicle ahead via Vehicle-to-Vehicle (V2V) communication) as input to control the steering angle. Our model demonstrates the lowest error when compared to the other existing approaches in the literature.

Three image analysis methods were studied and evaluated to solve the problem of removing long stems attached to mechanically harvested oranges : colour segmentation based on linear discriminant analysis , contour curvature analysis , and a thinning process which involves iterating until the stem becomes a skeleton . These techniques are able to determine the presence or absence of a stem with certainty , to locate the stems from random views with more than 90% accuracy and from profile images with an accuracy ranging from 92 и 4% to 100% depending on the method used . Finally , determination of the length and cutting point of the stem is achieved with only 3 и 8% of failures .÷ 1996 Silsoe Research Institute . IntroductionMechanical harvesting of citrus fruits brings some additional problems that have not been present in manual harvesting , such as the presence of fruits with long stems , with leaves or without calyx after detachment from the tree . Long stems and leaves can cause damage on adjacent fruits , while the absence of calyx opens a way for possible infections during transport and storage . This also means a loss of uniformity of the product which is not desirable for the fresh market . Therefore , a system for cutting long stems and for detecting the absence of calyx before the fruit arrives at the packing houses would be advantageous .Mechanical destemming systems , such as the one reported by Chen , 1 are usually based on random rotation of the oranges against cutting surfaces . However , the contact with these surfaces may cause some damage to the fruits . Some The random orientation of fruit on conveyor belts is often a problem in stem cutting systems . The sphericity of oranges hinders the possibility of mechanical orientation , so the detection of the stem -calyx area by a camera seems to be an adequate way to orientate the fruit . Once the stem has been situated at the correct position , it can be characterized and measured so the decision to cut the stem of f or not can be made .A system for orientation of oranges would be of interest for implementation on the CITRUS robot 6 in order to cut the stem of f after the picking operation .The objectives of this study were to design reliable image analysis methods (1) to locate , using colour vision , the stem -calyx area of oranges randomly presented to a camera , in order to be able to orientate the fruit , as well as to classify it on the basis of presence or absence of stem and leaves , and (2) to study the profiles of the fruit previously oriented and locate the stem to determine its length and cutting point . . Materials and methodsColour images were acquired with a charge coupled Since dif fuse light is highly ef fective in eliminating shadows and specular reflection , and in preserving well-defined edges , an illumination chamber with indirect fluorescent light and dif fusing material was built and used to take colour images , while illumination by contrast was employed to acquire profile images .The following three working methods were defi...

The rapid growth of connected and automated vehicle (CAV) solutions have made a significant impact on the safety of intelligent transportation systems. However, similar to any other emerging technology, thorough testing and evaluation studies are of paramount importance for the effectiveness of these solutions. Due to the safety-critical nature of this problem, large-scale real-world field tests do not seem to be a feasible and practical option. Thus, employing simulation and emulation approaches are preferred in the development phase of the safety-related applications in CAVs. Such methodologies not only mitigate the high cost of deploying large number of real vehicles, but also enable researchers to exhaustively perform repeatable tests in various scenarios. Software simulation of very large-scale vehicular scenarios is mostly a time consuming task and as a matter of fact, any simulation environment would include abstractions in order to model the real-world system. In contrast to the simulation-based solutions, network emulators are able to produce more realistic test environments. In this work, we propose a high-fidelity hardware-in-the-loop network emulator framework in order to create testing environments for vehicle-to-vehicle (V2V) communication. The proposed architecture is able to run in real-time fashion in contrast to other existing systems, which can potentially boost the development and validation of V2V systems.

No abstract

No abstract