Simulation is essential for developing robotic manipulation systems, particularly for task and motion planning (TAMP), where symbolic reasoning interfaces with geometric, kinematic, and physics-based execution. Recent advances in Large Language Models (LLMs) enable robots to generate symbolic plans from natural language, yet executing these plans in simulation often requires robot-specific engineering or planner-dependent integration. In this work, we present a unified pipeline that connects an LLM-based symbolic planner with the Kautham motion planning framework to achieve generalizable, robot-agnostic symbolic-to-geometric manipulation. Kautham provides ROS-compatible support for a wide range of industrial manipulators and offers geometric, kinodynamic, physics-driven, and constraint-based motion planning under a single interface. Our system converts language instructions into symbolic actions and computes and executes collision-free trajectories using any of Kautham's planners without additional coding. The result is a flexible and scalable tool for language-driven TAMP that is generalized across robots, planning modalities, and manipulation tasks.
This study presents a method for constructing metric depth maps based on relative depth maps for a 4-DOF manipulator with a single RGB camera, which is attached to an end effector. To create a relative depth map of a scene, we utilize the Depth Anything V2 model along with a system of twelve ArUco markers positioned at varying distances from one another. Unlike traditional methods for determining distances to objects using ArUco markers, the proposed approach does not require a continuous presence of markers within a camera's field of view; it only necessitates measuring a distance to a few markers in an initial frame. The proposed method enables the robot to perform object localization using a relative depth map concurrently with an object search process.
Global path planning is a crucial component in mobile robot navigation. To provide a good basis that is both feasible and yields fast motion, it should include both robot kinematics and collision avoidance. We introduce a new end-to-end algorithm to solve the mobile robot path planning problem for non-holonomic robots. We consider maximum steering angle, velocity, and acceleration constraints to optimize paths for both collision-avoidance and short tracking time. To this end, we employ B-Splines and metaheuristic optimization algorithms. Experiments show the suitability of our algorithm.
This paper addresses the challenge of determining the shortest path in a domain with static obstacles by leveraging the concepts of visibility graphs and visibility polygons. The core idea is that, since the obstacles are considered static, the only parts of the overall graph that are subject to change are the edges that connect the initial and final nodes to the rest of the graph. This allows the overall graph to be divided into two sub-graphs: (i) a static sub-graph that captures the visibility among subsets of the obstacle vertices and that can be computed offline, and (ii) a dynamic sub-graph that connects the initial and final points to the static sub-graph. This division entails fewer online computations when computing the shortest path, as building the entire visibility graph can be time-consuming. Additionally, we propose a purely geometrical approach based on visibility polygons to determine the connections between the initial and ending nodes. We then divide the environment into sub-regions to reduce the search space, making the dynamic linking of the starting and ending nodes faster.
This paper presents the design and implementation of a crutches-like compact ‘biped’ robot inspired by rehabilitative exoskeletons. Building on the compass-like biped concept, the proposed system is tailored for practical real-world use with only three actuators. The design allows the robot to perform both straight-line walking and steering maneuvers, emphasizing simplicity and cost-efficiency. Unactuated crutches are incorporated to enhance stability and reduce the load on the actuated leg during locomotion. Experimental results validate the robot's ability to navigate indoor environments with flat and regular surfaces, showcasing the practicality and robustness of the proposed design.
Currently, programming sequences for pneumatic industrial processes are carried out using different methods: pneumatic programming (in disuse), electro pneumatic (using electrical contactors) and programming with a PLC (Programmable Logic Controller). Among these, programming using Ladder Logic is the most commonly used and widely recognized for PLC devices. However, its implementation might take large periods of time, becoming a tedious task and increasing its level of complexity as a function of the number of cylinders, movements and phases that the process requires. This work develops and implements a control algorithm in the LabView graphical environment, which is subsequently connected to a Siemens PLC via an NI OPC (National Instruments OLE for Process Control) Server. The control algorithm is experimentally tested and proven successful in various scenarios of industrial pneumatic sequences.
Automated cargo transport is a frequent task in intralogistics and industrial scenarios, but is usually carried out by a single robot, thus limited to its payload capabilities. In this paper, we present a method for coordinated motion planning and control of two autonomous mobile robots for the joint transport of cargo that is too large for a single robot. We consider the two robots with a large cargo on top of them as an any-shape single kinematic chain with cargo. We plan an optimal collision free path for the entire differential drive kinematic chain that ensures the transport of the cargo to the desired target location. The implemented algorithm sends control commands to each robot separately, which achieves the coordinated execution of the planned path. Finally, the proposed method is validated in simulated and real-world scenarios.
Development and application of control strategies that regulate movements of prosthetic hands play a vital role in improving both operational efficiency and overall user satisfaction associated with these advanced devices. This paper presents a Sliding Mode Controller (SMC) that is based on the Timoshenko beam theory, specifically designed for a tendon-driven soft continuum wrist integrated with a prosthetic hand known as ‘PRISMA HAND II”. The controller utilizes the Timoshenko modeling technique to develop kinematic and dynamic models of the soft wrist. These models play a crucial role in the SMC to determine necessary tendon forces that are applied to the wrist to facilitate desired movements of the hand. The paper includes simulation studies and experimental tests that validate effectiveness of the controller in managing wrist movements.
Classical economic-emission dispatch considers only conventional generators. But the need of environment sustainability has promoted the penetration of renewable energy sources (RES). RES can help to provide clean energy at decreased generation cost. A major problem faced by generating plants is network reliability issues in power system arising from the peak hour's power demand. Application of demand side management (DSM) measures can help to ease the enormous peak hour's pressure in smart grid system. This paper presents a solution of peak hour's demand management with dynamic economic-emission dispatch (DEED). This paper provides a comparative analysis of the implications of deploying DSM strategies in smart grid system with/without integration of RES. The proposed system uses Grasshopper optimization algorithm to find the solutions. Optimal control of conventional generating units and energy storage system with application of DSM programs demonstrates the potential to diminish the peak overload and economic power dispatch with emission control.
This paper details the development of a digital twin for off-the-shelf optical tracking systems using CoppeliaSim, a robotics simulation platform, and Python scripting. The proposed simulation enables users to configure various parameters, such as camera position, lighting conditions, noise levels, and lens distortion, to mimic optical motion tracking in a controlled and customizable environment. By generating synthetic datasets, the tool offers a flexible approach to testing and refining motion capture setups prior to the physical installation of the system. The results highlight the effectiveness of virtual environments in reducing both the complexity and cost associated with the design phase of traditional motion capture systems.
This paper presents the validation of a smart ad hoc system designed to approximate the sea surface velocity field using measurements obtained from multiple floating sensors. Recognizing that point data alone cannot fully capture the domain's circulation, we propose a method that uses a simplified two-dimensional flow model as a surrogate for submesoscale flow, reconstructing the entire velocity field from the scattered data. To ensure reliable flow dynamics, the approach addresses the problem as a model-fitting process, adjusting the boundary conditions of numerical simulations to align the flow with measured data. Experimental validation was performed with the collected measurements, which involved the deployment of Global Positioning System (GPS) drifters in the Kvarner Bay area. The results highlight the system's adaptability to various domains, measurement types, and conditions, demonstrating its effectiveness for real-world submesoscale applications where an approximate velocity field is acceptable.
With recent advances in industrial robotics, autonomous robotic arms need to deal with complex environments using sensors such as depth cameras. Structured-light cameras are often employed for this purpose as they provide high precision measurements and increasing speed acquisition. Furthermore, as digital twins are becoming more common for training and guiding real robots, there is a growing need for high-performance sensor simulations. This paper thus introduces a fast, GPU-implemented, and realistic simulation tool for structured-light depth cameras. It can compute 3D point clouds with noisy coordinates in real-time for a set of cameras in complex, dynamic scenes simulated by a physics engine. It can also interactively evaluate coverage and overlap of simulated point clouds on objects of interest. Finally, by maximizing their values, as a pre-process or during simulation for motionless scenes, it allows optimal cameras positioning. To demonstrate the capabilities of our simulation tool, we present two use cases. The first showcases a real-time simulation of a high-speed structured-light camera operating at 20 frames per second, mounted on a robotic arm that dismantles an electrical battery. The second illustrates the optimal positioning of a set of fixed structured-light cameras in a cluttered environment, where a robotic arm performs the cutting of a glove box. To our knowledge, our tool is the first to offer such real-time simulations and rapid structured-light camera positioning in complex robotic environments.
The global shortage of specialized surgeons poses a significant challenge to providing equitable surgical care, particularly in underserved regions where geographic and professional barriers limit access. This study introduces an intuitive computer vision-based telesurgery system utilizing a standard webcam, enabling surgeons to perform remote operations without the need for bulky or high-cost equipment. Additionally, a motion transmission system for the Sawyer robot was developed to accurately replicate human movement trajectories. Experiments were developed based on basic surgical movements. The results demonstrated an absolute mean error of 2.36 mm horizontally and 2.49 mm vertically for linear motions, and 3.93 mm for circular trajectories. These values fall within the operational range of 2.2 mm to 5.5 mm established by prior studies. Compared to existing systems with high implementation costs, the proposed method offers a more accessible and cost-efficient solution, making telesurgery more viable in resource-constrained environments. These findings highlight the potential of the system to reduce training time and enhance movement accuracy, thereby fostering the adoption of robotic-assisted telesurgery in underserved areas.
Twin Delayed Deep Deterministic Policy Gradient (TD3) is a famous reinforcement learning algorithm which continues to generate state-of-the-art results since its introduction in 2018. In this article, we present a generalization of TD3, called Triple Delayed Deep Deterministic Policy Gradient (3D3), which introduces additional Q-critics and a flexible update function. We analyze its behavior for different specifications when applied to benchmark datasets and compare its performance to TD3 and another extension proposed in the literature. Moreover, we discuss its mathematical properties from a theoretical point of view, including estimation bias and convergence properties under discrete state-action space assumptions.
This paper delves into the domain of Brushless Direct Current (BLDC) Motors that have become significantly popular in modern day applications like automotive, aerospace, robotics (unmanned vehicles) etc. because of the edge they have over their brushed counterpart. Unlike brushed DC motors, which use brush and commutator segments for commutation, BLDC motors use electronic commutation which eliminates wear and tear of mechanical parts, frictional loss etc. This paper discusses mathematical modelling of a BLDC motor, presents the insights about design and implementation of Field Oriented Control (FOC), initial calculation of Proportional Integral Differential (PID) gains and final implementation of FOC algorithm for BLDC Motors of UUV. This technique is far superior than other traditional methods, like Trapezoidal control and sinusoidal control, and implements independent control of torque and flux with the help of various transforms. It is a very useful and precise technique which results in a superior dynamic response, less torque ripple and high efficiency. The key objective is to dispense a holistic approach for designing PID controller for a BLDC motor. The performance of the controller was tested for no load and load condition. The final results demonstrated the anticipated superior dynamic response.
Image captured from cameras can get blurred when it is mounted on rotating platform like gimbal, which is moving continuously at high speed and high acceleration. Scanning and surveillance application requires panoramic image or video to be generated from continuously rotating platform. The method employed here uses two motorized mirrors along optical axis, one for scanning the view mounted on a gimbal and another mounted on Fast steering mechanism for freezing the scene during the exposure time of the camera detector. The motion of gimbal and angular velocity of fast steering mechanism are synced in position and time with frame acquisition in camera to obtain a de-blurred image. Fast steering mirror (FSM) is used to compensate blurred image caused by scan motion of platform.
A Shear wall is the most appropriate structure shape in reinforced concrete buildings. Thus, examining the Shear Wall in structural systems is vital because it effectively resists lateral loads such as wind and seismic forces. The behaviour of such reinforced concrete building with Shear Wall differs from that of other reinforced concrete constructions. Thus, structural analysis of Shear Walls with openings for natural light, ventilation, entry, escape etc is required This factor has been analysed in terms of maximum storey displacement and base shear for model with Shear Walls and openings, using Equivalent static method to analyse the seismic performance of the structures. A comprehensive dataset is prepared using Staad Pro v8i software. The models are developed for the number of stories from 5 to 40, with different sizes of openings in the shear wall. A total of 240 models were analysed in III, IV and V zones of the earthquake, as defined by Indian code. The study involves the data set preparation and training, testing and validation using different machine learning techniques like deep neural networks, artificial neural networks and random forests. so that it will be helpful in predicting the seismic response of any similar multi-storey building in terms of Maximum storey displacement and Base shear.
This paper describes an innovative solution in the field of navigation systems, in particular for obstacle avoidance with Time-of-Flight (TOF) sensor(s). It is inspired by a “divide and conquer” like idea. Obstacle avoidance consists in choosing the correct robot rotation angle to avoid impact with (moving) objects. The solution used is based on a simple state machine based only on “turn” and “go straight” commands. The system resets the chosen direction each time the robot moves forward to avoid potential rotation loops and unpleasant rotations. The other innovation consists in dividing into $N$ vertical parts the TOF(s) we are considering and calculating separately a measure of each part. The combination of these values (the measure calculated on each part and chosen direction) will determine the output rotation to avoid the obstacle. Moreover, a peculiar preprocessing is done at the beginning to increase performance. The system works in real-time also in a dynamic environment and can be easily adapted to different TOF-like sensors.
This paper employs a massively parallel deep re-inforcement learning approach to model human behavior in a quadruped robot system designed to function as a guide dog for visually impaired individuals. The supervisory network operates alongside a control scheme that ensures the safe interaction between the human user and the robot. It distinguishes between several simplified human behaviors, including stopping, turning, spinning, and walking straight. The trained network builds upon a low-level control network that effectively manages the quadruped's locomotion across challenging terrains. The effectiveness of the proposed approach is validated in a physics-engine-based simulation environment.
Dual-arm manipulation is a key enabler for significantly enhancing the interaction between humans and robots, and their capabilities to purposefully shape the surrounding environment. However, the spatiotemporal coordination between the motion of the hands required for this type of actions makes their planning not trivial. A proper definition of these coordination patterns moving from the human example could simplify their translation on the robot side, fostering the generation of effective bimanual tasks. In this work, we propose Multivariate functional Principal Component Analysis (MfPCA) as a mathematical tool to encode inter-hands temporal kinematic covariations in terms of principal spatiotemporal coordination patterns in the Cartesian domain. We compared these patterns extracted from a dataset of human bimanual tasks with those resulting from the usage of classical fPCA, applied independently to each hand (univariate fPCA). We found that MfPCA allows for a better classification of the tasks, with respect to a state of the art taxonomy. For what concerns motion planning, MfPCA and fPCA yield similar accuracy in the reconstruction of the motion, but with a smaller number of principal components needed in the MfPCA case. These results, although preliminary, can open interesting perspectives for the usage of MfPCA for human-like bimanual motion planning and control of robotic manipulators, as well as for action recognition, to enable a more effective human-robot interaction.
