Development of a Particle Filter-Based Path Tracking Algorithm of Autonomous Trucks with a Single Steering and Driving Module Using a Monocular Camera.

Recently, in various fields, research into the path tracking of autonomous vehicles and automated guided vehicles has been conducted to improve worker safety, convenience, and work efficiency. For path tracking of various systems applied to autonomous driving technology, it is necessary to recognize the surrounding environment, determine technology accordingly, and develop control methods. Various sensors and artificial-intelligence-based perception methods have limitations in that they must learn a large amount of data. Therefore, a particle-filter-based path tracking algorithm using a monocular camera was used for the recognition of target RGB. The path tracking errors were calculated and a linear-quadratic-regulator-based desired steering angle were derived. The autonomous trucks were steered and driven using a pulse-width-modulation-based steering and driving motor. Based on an autonomous truck with a single steering and driving module, it was verified that the path tracking could be used in three evaluation scenarios. To compare the LQR-based path tracking control performance proposed in this paper, an elliptical path tracking scenario using a conventional sliding mode control with robust control performance was performed. The results show that the RMS of the lateral preview error of the SMC was approximately 18% larger than that of the LQR-based method.

Surface flattening of the human left atrium and proof-of-concept clinical applications.

Surface flattening in medical imaging has seen widespread use in neurology and more recently in cardiology to describe the left ventricle using the bull's-eye plot. The method is particularly useful to standardize the display of functional information derived from medical imaging and catheter-based measurements. We hypothesized that a similar approach could be possible for the more complex shape of the left atrium (LA) and that the surface flattening could be useful for the management of patients with atrial fibrillation (AF). We implemented an existing surface mesh parameterization approach to flatten and unfold 3D LA models. Mapping errors going from 2D to 3D and the inverse were investigated both qualitatively and quantitatively using synthetic data of regular shapes and computer tomography scans of an anthropomorphic phantom. Testing of the approach was carried out using data from 14 patients undergoing ablation treatment for AF. 3D LA meshes were obtained from magnetic resonance imaging and electroanatomical mapping systems. These were unfolded using the developed approach and used to demonstrate proof-of-concept applications, such as the display of scar information, electrical information and catheter position. The work carried out shows that the unfolding of complex cardiac structures, such as the LA, is feasible and has several potential clinical uses for the management of patients with AF.