Zoekresultaten

Robots and moral obligations

From the article: Using Roger Crisp’s arguments for well-being as the ultimate source of moral reasoning, this paper argues that there are no ultimate, non-derivative reasons to program robots with moral concepts such as moral obligation, morally wrong or morally right. Although these moral concepts should not be used to program robots, they are not to be abandoned by humans since there are still reasons to keep using them, namely: as an assessment of the agent, to take a stand or to motivate and reinforce behaviour. Because robots are completely rational agents they don’t need these additional motivations, they can suffice with a concept of what promotes well-being. How a robot knows which action promotes well-being to the greatest degree is still up for debate, but a combination of top-down and bottom-up approaches seem to be the best way. The final publication is available at IOS Press through http://dx.doi.org/10.3233/978-1-61499-708-5-184

Robotics Simulation Labs

Deep Reinforcement Learning Applied to a Robotic Pick-and-Place Application

Industrial robot manipulators are widely used for repetitive applications that require high precision, like pick-and-place. In many cases, the movements of industrial robot manipulators are hard-coded or manually defined, and need to be adjusted if the objects being manipulated change position. To increase flexibility, an industrial robot should be able to adjust its configuration in order to grasp objects in variable/unknown positions. This can be achieved by off-the-shelf vision-based solutions, but most require prior knowledge about each object tobe manipulated. To address this issue, this work presents a ROS-based deep reinforcement learning solution to robotic grasping for a Collaborative Robot (Cobot) using a depth camera. The solution uses deep Q-learning to process the color and depth images and generate a greedy policy used to define the robot action. The Q-values are estimated using Convolutional Neural Network (CNN) based on pre-trained models for feature extraction. Experiments were carried out in a simulated environment to compare the performance of four different pre-trained CNNmodels (RexNext, MobileNet, MNASNet and DenseNet). Results showthat the best performance in our application was reached by MobileNet,with an average of 84 % accuracy after training in simulated environment.

What pupils can learn from working with robotic direct manipulation environments

This study investigates what pupils aged 10-12 can learn from working with robots, assuming that understanding robotics is a sign of technological literacy. We conducted cognitive and conceptual analysis to develop a frame of reference for determining pupils' understanding of robotics. Four perspectives were distinguished with increasing sophistication; psychological, technological, function, and controlled system. Using Lego Mindstorms NXT robots, as an example of a Direct Manipulation Environment, we developed and conducted a lesson plan to investigate pupils' reasoning patterns. There is ample evidence that pupils have little difficulty in understanding that robots are man-made technological and functional artifacts. Pupils' understanding of the controlled system concept, more specifically the complex sense-reason-act loop that is characteristic of robotics, can be fostered by means of problem solving tasks. The results are discussed with respect to pupils' developing technological literacy and the possibilities for teaching and learning in primary education.

Robotic Leakage Detection

Within our research on robotic gas detection, we have focused on making a prototype based on Boston Dynamics SPOT, because it takes a lot of difficulties out of prototyping. For instance, it has its own obstacle avoidance algorithm, good drivers are available for ROS2, and SPOT is meant for outdoor navigation. Being a legged robot means that it can easily traverse curbs, shrubberies, unstable soil and even stairs. For this document, we are going to use the insights that we used when looking for a solution for SPOT.

Design-to-Robotic-Production and -Assembly for Architectural Hybrid Structures

The Design-to-Robotic-Production and -Assembly (D2RP&A) process developed at Delft University of Technology (DUT) has been scaled up to building size by prototyping of-site a 3.30 m high fragment of a larger spaceframe structure The fragment consists of wooden linear elements connected to a polymer node printed at 3D Robot Printing and panels robotically milled at Amsterdam University of Applied Science (AUAS). It has been evaluated for suitability for assembly on-site without temporary support while relying on human-robot collaboration. The constructed architectural hybrid structure is proof of concept for an on- and off-site D2RP&A approach that is envisioned to be implemented using a range of robots able to possibly address all phases of construction in the future.