Autonomous driving at unsignalized intersections is still considered a challenging application for machine learning due to the complications associated with handling complex multi-agent scenarios characterized by a high degree of uncertainty. Automating the decision-making process in these safety-critical environments involves comprehending multiple levels of abstraction associated with learning robust driving behaviors to enable the vehicle to navigate efficiently. In this survey, we aim at exploring the state-of-the-art techniques implemented for decision-making applications, with a focus on algorithms that combine Reinforcement Learning (RL) and deep learning for learning traversing policies at unsignalized intersections. The reviewed schemes vary in the proposed driving scenario, in the assumptions made for the used intersection model, in the tackled challenges, and in the learning algorithms that are used. We have presented comparisons for these techniques to highlight their limitations and strengths. Based on our in-depth investigation, it can be discerned that a robust decision-making scheme for navigating real-world unsignalized intersection has yet to be developed. Along with our analysis and discussion, we recommend potential research directions encouraging the interested players to tackle the highlighted challenges. By adhering to our recommendations, decision-making architectures that are both non-overcautious and safe, yet feasible, can be trained and validated in real-world unsignalized intersections environments.

In multiagent coordination tasks, motion trajectories are required to satisfy a range of constraints that present significant implementation challenges due to the limited onboard sensing and communication capacities. This article introduces distributed control laws that integrate nonholonomic motion constraints into bearing-based designs to enable time-varying formation tracking with minimal onboard resources. Unlike state-of-the-art formation control solutions, this approach maintains formation shape through relative bearing feedback and orientation alignment rather than tracking global target locations or regulating interagent relative positions and velocities. This distributed controller design has been validated in two deployment scenarios: 1) leaderless nonhierarchical formations and 2) leader-follower hierarchical formations. In hierarchical formations, follower agents employ a speed estimator within the orientation alignment framework to reach velocity consensus with the leader agent. The proposed controllers guarantee accurate tracking of time-varying reference trajectories, preserve the desired formation structures, and achieve velocity consensus for both nonhierarchical and hierarchical formations, as established by analysis and validated through simulations and experiments.

When applied to autonomous vehicle (AV) settings, action recognition can enhance an environment model's situational awareness. This is especially prevalent in scenarios where traditional geometric descriptions and heuristics in AVs are insufficient. However, action recognition has traditionally been studied for humans, and its limited adaptability to noisy, un-clipped, un-pampered, raw RGB data has limited its application in other fields. To push for the advancement and adoption of action recognition into AVs, this work proposes a novel two-stage action recognition system, termed RALACs. RALACs formulates the problem of action recognition for road scenes, and bridges the gap between it and the established field of human action recognition. This work shows how attention layers can be useful for encoding the relations across agents, and stresses how such a scheme can be class-agnostic. Furthermore, to address the dynamic nature of agents on the road, RALACs constructs a novel approach to adapting Region of Interest (ROI) alignment to agent tracks for downstream action classification. Finally, our scheme also considers the problem of active agent detection, and utilizes a novel application of fusing optical flow maps to discern relevant agents in a road scene. We show that our proposed scheme can outperform the baseline on the ICCV2021 Road Challenge dataset (Singh et al., 2023) algorithm and by deploying it on a real vehicle platform, we provide preliminary insight to the usefulness of action recognition in decision making. The code is publicly available at https://github.com/WATonomous/action-classification.

Trajectory planning and control have historically been separated into two modules in automated driving stacks. Trajectory planning focuses on higher-level tasks like avoiding obstacles and staying on the road surface, whereas the controller tries its best to follow an ever changing reference trajectory. We argue that this separation is (1) flawed due to the mismatch between planned trajectories and what the controller can feasibly execute, and (2) unnecessary due to the flexibility of the model predictive control (MPC) paradigm. Instead, in this paper, we present a unified MPC-based trajectory planning and control scheme that guarantees feasibility with respect to road boundaries, the static and dynamic environment, and enforces passenger comfort constraints. The scheme is evaluated rigorously in a variety of scenarios focused on proving the effectiveness of the optimal control problem (OCP) design and real-time solution methods. The prototype code will be released at github.com/WATonomous/control.

Reinforcement learning-based techniques, empowered by deep-structured neural nets, have demonstrated superiority over rule-based methods in terms of making high-level behavioral decisions due to qualities related to handling large state spaces. Nonetheless, their training time, sample efficiency and the feasibility of the learnt behaviors remain key concerns. In this paper, we propose a novel hierarchical reinforcement learning-based decision-making architecture for learning left-turn policies at unsignalized intersections with feasibility guarantees. The proposed technique is comprised of two layers; a high-level learning-based behavioral planning layer which adopts soft actor-critic (SAC) principles to learn high-level, non-conservative yet safe, driving behaviors, and a low-level motion planning layer that uses Model Predictive Control (MPC) framework to ensure feasibility of the two-dimensional left-turn maneuver. The high-level layer generates reference signals of velocity and yaw angles for the ego vehicle taking into account safety and collision avoidance with the intersection vehicles, whereas the low-level motion planning layer solves an optimization problem to track these reference commands taking into account several vehicle dynamic constraints and ride comfort. While training the behavioral SAC-based planning layer, We develop an adaptive entropy regularization technique that results in faster convergence, higher mean rewards, and lower collision rates. We validate the proposed decision-making scheme in simulated environments and compare with other model free Reinforcement Learning (RL) baselines. The results demonstrate that the proposed integrated framework possesses better training and navigation capabilities.

This paper presents a path following control designed for human-robot collaborative tasks. The path following strategy allows the task description and human-robot interaction to be defined according to components tangent and transversal to a nominal path. With this formulation, the control objectives for each task can be customized to generate motions towards task completion and getting on the nominal path when switching between tasks. Simulation examples of the path following control is presented, demonstrating the ability for the proposed controllers to be customized according to objectives of the human-robot collaboration.

Environment modeling is the backbone of how autonomous agents understand the world, and therefore has significant implications for decision-making and verification. Motivated by the success of relational mapping tools such as Lanelet2, we present the Dynamic Relation Graph (DRG). The DRG is a novel method for extending prior relational maps to include online observations, creating a unified en-vironment model which incorporates both prior and online data sources. Our prototype implementation models a finite set of heterogeneous features including road signage and pedestrian movement. However, the methodology behind the DRG can be expanded to a wider range of features in a fashion that does not increase the complexity of behavioral planning. Simulated stress tests indicate the DRG's effectiveness in decreasing decision-making complexity, and deployment on the University of Waterloo's WATonomous research vehicle demonstrates its practical utility. The prototype code will be released at github.com/WATonomous/DRG.

While supervised detection and classification frameworks in autonomous driving require large labelled datasets to converge, Unsupervised Domain Adaptation (UDA) approaches, facilitated by synthetic data generated from photoreal simulated environments, are considered low-cost and less time-consuming solutions. In this paper, we propose UDA schemes using adversarial discriminative and generative methods for lane detection and classification applications in autonomous driving. We also present Simulanes dataset generator to create a synthetic dataset that is naturalistic utilizing CARLA’s vast traffic scenarios and weather conditions. The proposed UDA frameworks take the synthesized dataset with labels as the source domain, whereas the target domain is the unlabelled real-world data. Using adversarial generative and feature discriminators, the learnt models are tuned to predict the lane location and class in the target domain. The proposed techniques are evaluated using both real-world and our synthetic datasets. The results manifest that the proposed methods have shown superiority over other baseline schemes in terms of detection and classification accuracy and consistency. The ablation study reveals that the size of the simulation dataset plays important roles in the classification performance of the proposed methods. Our UDA frameworks are available at github.com/anita-hu/sim2real-lane-detection and our dataset generator is released at github.com/anita-hu/simulanes.

In this paper, a new approach for lane change and double-lane change planning and following for autonomous driving is proposed. Herein, we introduce a novel technique, based on exponential functions, to generate online and feasible lane change maneuvers; these maneuvers satisfy the constraints on the maximum allowable curvature a given vehicle can handle. In addition, a simultaneous local path planning and path-following control framework is adapted. The framework utilizes a multi-threading architecture to run the local planner module concurrently with the control module. The planning module generates parametric reference paths based on the proposed lane change and double-lane change maneuvers. The control module is based on a Model Predictive Path-following Control (MPFC) scheme, which achieves the path following objective while satisfying vehicle’s state and control limits. To validate the proposed framework, several real-time simulation scenarios are designed and tested on CARLA soft real-time vehicle simulator. The results show the effectiveness of the proposed framework in generating and smoothly following lane change maneuvers.

Autodriver algorithm aims to develop a path-following algorithm for autonomous vehicles using road geometry data and vehicle dynamics. In this study, a novel smart Autodriver algorithm is developed according to practical implications utilizing a more realistic vehicle model and consideration of real-time applicability. A ghost-car path-following approach is introduced to define the desired location of the vehicle at every instance during various maneuvers. Key steady-state characteristics of turning vehicles, namely the curvature, yaw rate, and side-slip responses are discussed and used to construct a path-following controller based on the Autodriver algorithm. A feedback control based on Sliding Mode Control (SMC) is also designed and applied to minimize transient errors between the road and the vehicle positions. Finally, simulations are performed to analyze the path-following performance of the proposed scheme compared to a Model Predictive Controller (MPC) as a widely accepted popular method for autonomous vehicles. Hardware-in-the-loop (HIL) tests are also performed to investigate real-time applicability of the controllers. The results show promising controller performance in terms of error minimization, passenger comfort, and low computational cost for the proposed method.

This article presents redundancy resolution and disturbance rejection via torque optimization in hybrid cable-driven robots (HCDRs). To begin with, we present a redundant HCDR for nonlinear whole-body system modeling and model reduction. Based on the reduced dynamic model, two new methods are proposed to solve the redundancy resolution problem: 1) joint-space torque optimization for actuated joints (TOAJ) and 2) joint-space torque optimization for actuated and unactuated joints (TOAUJ), and they can be extended to other HCDRs. Compared to the existing approaches, this article provides the first solution (TOAUJ-based method) for HCDRs that can solve the redundancy resolution problem as well as disturbance rejection. Additionally, this article develops detailed algorithms targeting TOAJ and TOAUJ implementation. A simple yet effective controller is designed for analysis and validation. Case studies are conducted to evaluate the performance of TOAJ and TOAUJ. The results show the effectiveness of the aforementioned approaches.

Deployment of unmanned aerial vehicles (UAVs) as aerial base stations to support cellular networks can deliver a fast and flexible solution for serving high and varying traffic demand. In order to adequately leverage the benefit of UAVs deployment, their efficient placement is of utmost importance, and requires to intelligently adapt to the environment changes. In this paper, we propose novel learning-based mechanisms for the three-dimensional deployment of UAVs assisting terrestrial networks in the downlink for overloaded situations. The problem is modeled as a game among UAVs. To solve the game, we utilize tools from reinforcement learning, and develop low complexity algorithms based on the multi-armed bandit and satisfaction methods to learn UAVs' locations. Simulation results reveal that the proposed satisfaction based UAV placement algorithm can yield significant performance gains up to about 50% and 41% in terms of throughput and the number of outage users, respectively, compared to a learning based benchmark algorithm.

In this paper, a novel soft inflatable arm is proposed for telepresence robots. The new proposed structure of the arm is achieved by a very common and low-cost inflatable material and it is very light, weighing only about 50 g. However, it can realize agile movement by driving six tiny cables installed in the shoulder and elbow joints. The soft inflatable arm can work by pumping air at a very low pressure (7.32 ± 3.45 kPa) and allows direct and soft human contact without any external force sensors. This paper proposes joint compressed models, joint kinematic models, and kinematic models for the whole inflatable robot arm. These models can be easily applied to multijoint arms. In addition, a new redundancy resolution method is also developed for the inflatable arm, which makes it easier to control resolution and is less complex than other traditional approaches. Numerous experiments have been conducted, including performances of accuracy, repeatability, motion trajectories, human-safe interaction, and remote interaction. Results are satisfactory and validate the expected performance of the proposed robotic arm.

Scheduling of multiple resource systems with resource constraints is a computationally expensive task especially when one is dealing with a large number of procedures. Scheduling through assessing the specific order of tasks for each procedure not only minimizes the makespan of a schedule but also maximizes the utilization of resources. This paper develops a new scheduling formulation that is computationally efficient and enforces constraints of multiple resource procedures. We showed that by applying genetic algorithm to this method we can schedule 100 of multiple resource procedures in less than 2 minutes on a personal computer. We also developed a dynamic scheduling technique that is capable of re-scheduling as new procedures arrive in a system. The scheduling performance is consistent with different numbers of procedures.

The huge amount of data generated by modern medicine has motivated us to develop decision support systems for improving health care applications. In this paper, we address the problem of clinical disease prediction given patient-reported symptoms and medical signs where patient records lack of semantic code annotation. We propose a novel context-enhanced disease prediction approach based on leveraging semantic and contextual medical entity relations. We have already exploited semantic relations of medical terminology for patient records search [2] but they were never considered for disease prediction in the literature. Patient signs and symptoms are first mapped to SNOMED-CT concepts, which compose a feature space for disease prediction. Our major contributions in this paper consist of expanding the feature space using semantic and contextual concept relations of SNOMED-CT. Based on patient's reported signs and symptoms, we use biomedical text mining tool, namely Metamap [1] to extract concepts of the SNOMED-CT metathesaurus. A “concept” in SNOMED-CT is a clinical meaning identified by a unique numeric identifier (ConceptId) and described via a set of words. For each concept, we define a medical entity context by integrating “defining” and “qualitative” medical aspects through the use of different types of semantic and contextual relationships of SNOMED-CT. Figure 1 illustrates the concept “Pneumonia” and its relations to other concepts.

This paper presents a laser scanner based reverse engineering (RE) system that generates accurate 3D solid models of bottles. The proposed system is designed for ease of use and has an open architecture, making it an ideal solution for bottle conveying system manufacturers. Conveyor system manufactures need a versatile RE system that can create 3D models of bottles to develop conveyor systems for their clients. This paper also presents a simple, robust filtering technique that can effectively remove the noise from the scanned laser data, yet retaining the shape of the scanned bottle. An error analysis is performed to compare our technique against another common data filtering technique, and sample 3D models are generated to contrast the effects of the different filtering approaches.

In this paper, we present a Bayesian-based Personalized Laboratory Tests prediction (BPLT) model to solve a real world medical problem: how to recommend laboratory tests to a group of patients? Given a patient who has conducted several laboratory tests, BPLT model recommends further laboratory tests that are the most related to this patient. We regard this laboratory test prediction problem as a special classification problem, where a new laboratory test belongs to either a "taken" or "not-taken" class. Our goal is to find the laboratory tests with high probability of "taken" and low probability of "not taken". Based on Bayesian method, the BPLT model builds a weighting function to investigate the correlations among laboratory tests and generate the rank of laboratory tests. In order to evaluate the proposed BPLT model, we further propose a novel evaluation metric to subjectively measure the accuracy of BPLT model. Experimental results show that BPLT model achieves good performance on the real data sets and provides a good solution to our real world application.

Type-2 fuzzy logic systems have recently been utilized in many control processes due to their ability to model uncertainties. This paper proposes a novel inference mechanism for an interval type-2 Takagi-Sugeno-Kang fuzzy logic control system (IT2 TSK FLCS) when antecedents are type-2 fuzzy sets and consequents are crisp numbers (A2-C0). The proposed inference mechanism has a closed form which makes it more feasible to analyze the stability of this FLCS. This paper focuses on control applications for the following cases: 1) Both plant and controller use A2-C0 TSK models, and 2) the plant uses type-1 Takagi-Sugeno (TS) and the controller uses IT2 TS models. In both cases, sufficient stability conditions for the stability of the closed-loop system are derived. Furthermore, novel linear-matrix-inequality-based algorithms are developed for satisfying the stability conditions. Numerical analyses are included which validate the effectiveness of the new inference methods. Case studies reveal that an IT2 TS FLCS using the proposed inference engine clearly outperforms its type-1 TSK counterpart. Moreover, due to the simple nature of the proposed inference engine, it is easy to implement in real-time control systems. The methods presented in this paper lay the mathematical foundations for analyzing the stability and facilitating the design of stabilizing controllers of IT2 TSK FLCSs and IT2 TS FLCSs with significantly improved performance over type-1 approaches.

Interval type-2 fuzzy logic control systems (IT2 FLCSs) have the potential of handling uncertainties better than type-1 FLCSs. However, lack of systematic design methodology of IT2 FLCSs limits their utility. This paper presents systematic methods to design interval IT2 Takagi-Sugeno-Kang (TSK) FLCSs that are PD-type and PI-type fuzzy controllers to satisfy certain desired transient response. We adopt the MacVicar-Whelan rule-base system and present general schemes for the design of IT2 TSK FLCSs, that include the design of the TSK consequent parameters. To validate the performance of the proposed controllers, some nonlinear plants have been considered. Results show that the IT2 TSK FLCSs satisfy the desired performance measures in terms of a set point tracking. Moreover, they reveal remarkable improvements in comparison to their type-1 counterparts for the plants considered in this paper.

Abstract: This paper describes a novel approach to implementation of a medical diagnosis expert system that can assist physicians with their daily practices. Differential artificial intelligence techniques are incorporated into a multi‐stage expert system to best represent the various phases of the patient diagnosis process. A weighted scoring system is used to represent the subjective analysis stage, while a rule‐based fuzzy expert system is employed to both interpret laboratory tests and imaging findings and suggest the final diagnosis. A model of various patient flow scenarios is presented to demonstrate the functionality of the proposed expert system. An actual example of patient walkthrough is used to demonstrate various computation steps from recording the patient chief complaint to arriving at the final diagnosis. It is shown that the conclusion arrived at by using the proposed system is consistent with a common diagnosis of a third party specialist who is asked to evaluate the performance of the system.

Type-2 fuzzy systems have successfully been applied in control applications. Due to the complicated structure of type-2 systems, they lack systematic control design and hence the stability of the system is not guaranteed. This paper presents stability analysis of dynamic type-2 Takagi-Sugeno-Kang (TSK) fuzzy systems. Novel inference mechanisms for type-2 TSK systems for the case when antecedents are type-2 and consequents are crisp numbers (A2-C0) are developed and utilized in fuzzy model generation. Owing to the simple nature of the proposed methods, they are easy to implement in real-time applications. One of the proposed inference mechanisms is used and the sufficient stability conditions for these systems are derived. It is shown that the criteria obtained herein must satisfy some linear matrix inequalities (LMI) and an algorithm is also presented to solve the obtained LMI. Two numerical examples are provided that detail the design method. The methodology presented proves to be an efficient approach to systematically design stable dynamic type-2 TSK fuzzy systems.

In this paper, automated micro-sized objects manipulation is investigated. The novelty of the proposed method lies on the compensation of all the nonlinear scaling forces which are dominant over gravitational force. A dynamic neural network has been added to a PD conventional controller for automated micromanipulation control. Weight-updating rules have been obtained in such a way that the system is uniformly ultimately bounded (UUB) in the sense of Lyapunov. Simulation results for controlled pushing of a micro-object have been illustrated and the efficiency of the method has been shown by comparing its result with that of a linear controller.

Wheeled-mobile robots operating in human environments typically encounter small steps. Surmounting steps is normally not considered when determining peak torque needs, yet it can be the maximum requirement. This work looks at the statics and dynamics of this situation to determine the necessary peak torque. It finds that using a dynamic model that includes the wheel elasticity is essential for properly representing a real-world tire. When torque is increased using a step function, energy is stored in the tire - higher tire elasticity eases climbing. Knowledge of this phenomenon could facilitate the use of smaller actuators. The model is numerically integrated and results are found to agree with experiment.

Inverse kinematics is a nonlinear problem that may have multiple solutions. A Genetic Algorithm(GA) for solving the inverse kinematics of a serial robotic manipulator is presented. The algorithm is capable of finding multiple solutions of the inverse kinematics through niching methods. Despite the fact that the number and position of solutions in the search space depends on the the position and orientation of the end-effector as well as the configuration of the robot, the number of GA parameters that must be set by a user are limited to a minimum through the use of an adaptive niching method. The only requirement of the algorithm is the forward kinematics equations which can be easily obtained from the link parameters and joint variables of the robot. For identifying and processing the outputs of this GA, a modified filtering and clustering phase is also added to the algorithm. The algorithm was tested to solve the inverse kinematic problem of a 3 degree-of-freedom(DOF) robotic manipulator.

In this paper, a neural network-based symmetric cipher design methodology is proposed to provide high performance data encryption. The proposed approach is a novel attempt to apply the parallel processing capability of neural networks for cryptography purposes. By incorporating neural networks approach, the proposed cipher releases the constraint on the length of the secret key. The proposed cipher is robust in resisting different cryptanalysis attacks and provides efficient data integrity and authentication services. The design of the symmetric cipher is presented and its security is analyzed. Simulation results are presented to validate the effectiveness of the proposed cipher design.

This paper proposes a novel method to compensate for the friction and all the scaling micro forces which add to the friction force in micromanipulation and more specifically in micro object pushing using an artificial neural network. The requirements for developing the neural network are addressed and the structure of the corresponding experimental setup is described in details.

This paper describes the development of a medical diagnosis expert system that can be used by physicians in their daily practices. Differential artificial intelligence techniques are incorporated into the expert system to best represent the various stages of the diagnosis process. A linear scoring system is used to represent the initial subjective analysis stage, while a rule-based fuzzy expert system is used to interpret lab tests and imaging studies to confirm final diagnosis. An actual example of patient walkthrough is used to demonstrate various computation steps from embedding the patient information to reaching the final diagnosis.

Let V=(n, k:) be an irreducible cyclic code over F/sub q//sup m/, the finite field of q/sup m/ elements. Let /spl alpha/_ be a basis of F/sub q//sup m/ over F/sub q/. Under the simplifying assumption (n,q)=1, it is shown that d/spl alpha/_(V), the q-ary image of V with respect to /spl alpha/, is decomposable into the direct sum of a fixed number of irreducible quasicyclic codes. This characterization allow us to obtain a lower bound on the minimum distance of d/spl alpha/_(V) by determining a lower bound on the number of not-null blocks in a typical codeword of d/spl alpha/_(V), for four specific subclasses of codes.