Browse Topic: Robotics

Items (458)

2021-01-00824/6/2021

This work presents a method to estimate the human body orientation using 2D images from a person view; the challenge comes from the variety of human body poses and appearances. The method utilizes OpenPose neural network as a human pose detector module and depth sensing module. The modules work together to extract the body orientation from 2D stereo images. OpenPose is proven to be efficient in detecting human body joints, defined by COCO dataset, OpenPose can detect the visible body joints without being affected by backgrounds or other challenging factors. Adding the depth data for each point can produce rich information to the process of 3D construction for the detected humans. This 3D point’s setup can tell more about the body orientation and walking direction for example. The depth module used in this work is the ZED camera stereo system which uses CUDA for high performance depth computation. One of the possible applications for this method is for social robots where the robot has to navigate between crowds; the human body orientation can be an important input for the path planner here. Other application might require the robot to face the human user for interaction, this method provides the robot with the info required to face the human user. This method is aimed for indoors activity to ensure higher accuracy.

Abughalieh, Karam, Alawneh, Shadi

Development of a Non-Parametric Robot Calibration Method to Improve Drilling Accuracy

2021-01-00033/2/2021

The drilling of large quantities of repetitive holes during the manufacture of large aerospace components is often considered a key limiting factor with regards to production efficiency. Whilst the desire within aerospace is to use relatively cheap six axis robot arms with drilling end effector units, their poor accuracy remains an obstacle. Robot calibration presents a way of improving robot accuracy such that aerospace drilling tolerances can be met, without permanently committing metrology equipment to an automation cell during production. Extensive research has been conducted into robot calibration by correcting the kinematic model, known as parametric calibration. This method is highly complex, and calibrates the robot across the entire working volume. This is often not required in industrial drilling applications, as drilling routines are often contained within a smaller volume of the robot reach. In this paper, a non-parametric method of robot calibration is proposed. This method involves calibrating within regions of the working volume where the robot pose is similar, and thus the effects of geometric errors in the kinematic model are roughly constant. By establishing the average positional error for each region, the accuracy can be locally improved by compensation through definition of the tool centre point. The proposed method can be completed without the use of kinematic models or complex mathematics, making it more suitable to industrial users. From experimental trials, a significant improvement in the positional accuracy of holes drilled using a standard six axis robot is reported, from 2 mm to 0.1 mm, well within the requirements of the majority of aerospace applications.

Scraggs, Chris, Smith, Thomas, Sawyer, Daniela, Davis, Matthew

Deep-Learned E-Skin Decodes Complex Human Motion

TBMG-37351

8/1/2020

A deep-learning powered single-strained electronic skin sensor can capture human motion from a distance. The single strain sensor placed on the wrist decodes complex five-finger motions in real time with a virtual 3D hand that mirrors the original motions. The deep neural network boosted by rapid situation learning (RSL) ensures stable operation regardless of its position on the surface of the skin.