Determining Exception Context in Assembly Operations from Multimodal Data.

Robot assembly tasks can fail due to unpredictable errors and can only continue with the manual intervention of a human operator. Recently, we proposed an exception strategy learning framework based on statistical learning and context determination, which can successfully resolve such situations. This paper deals with context determination from multimodal data, which is the key component of our framework. We propose a novel approach to generate unified low-dimensional context descriptions based on image and force-torque data. For this purpose, we combine a state-of-the-art neural network model for image segmentation and contact point estimation using force-torque measurements. An ensemble of decision trees is used to combine features from the two modalities. To validate the proposed approach, we have collected datasets of deliberately induced insertion failures both for the classic peg-in-hole insertion task and for an industrially relevant task of car starter assembly. We demonstrate that the proposed approach generates reliable low-dimensional descriptors, suitable as queries necessary in statistical learning.

Wide Skis As a Potential Knee Injury Risk Factor in Alpine Skiing.

Alpine skis with wider waist widths have recently become more popular. With such skis, the contact point of the ground reaction force during ski turns is displaced more medially from beneath the sole of the outer ski, which may present an increased risk of injury. The aim of this study was to investigate knee joint kinetics, kinematics, and lower limb muscle activation as a function of changes of the ski waist width in a laboratory setting. A custom skiing simulator was constructed to enable simulation of different ski waist widths in a quasi-static ski turn position. An optical system was used for capturing knee joint kinematics of the outer leg, whereas a force plate was used to determine the ground reaction force vector. The combination of both systems enabled values for external torques acting on the knee joint to be calculated, whereas electromyographic measurements enabled an analysis of knee flexor muscle activation. With respect to the outer ski, the knee joint external torques were independent of ski waist width, whereas knee joint external rotation and biceps femoris activation increased significantly with the increase of the ski waist width. Skier muscle and kinematics adaptation most probably took place to diminish the external knee joint torque changes when the waist width of the ski was increased. The laboratory results suggest that using skis with large waist widths on hard, frozen surfaces may change the load of knee joint surfaces. However, future research is needed to clarify if this may result in the increased risk of knee injury.

The Waist Width of Skis Influences the Kinematics of the Knee Joint in Alpine Skiing.

Recently alpine skis with a wider waist width, which medially shifts the contact between the ski edge and the snow while turning, have appeared on the market. The aim of this study was to determine the knee joint kinematics during turning while using skis of different waist widths (65mm, 88mm, 110mm). Six highly skilled skiers performed ten turns on a predefined course (similar to a giant slalom course). The relation of femur and tibia in the sagital, frontal and coronal planes was captured by using an inertial motion capture suit, and Global Navigation Satellite System was used to determine the skiers' trajectories. With respect of the outer ski the knee joint flexion, internal rotation and abduction significantly decreased with the increase of the ski waist width for the greatest part of the ski turn. The greatest abduction with the narrow ski and the greatest external rotation (lowest internal rotation) with the wide ski are probably the reflection of two different strategies of coping the biomechanical requirements in the ski turn. These changes in knee kinematics were most probably due to an active adaptation of the skier to the changed biomechanical conditions using wider skis. The results indicated that using skis with large waist widths on hard, frozen surfaces could bring the knee joint unfavorably closer to the end of the range of motion in transversal and frontal planes as well as potentially increasing the risk of degenerative knee injuries. Key pointsThe change in the skis' waist width caused a change in the knee joint movement strategies, which had a tendency to adapt the skier to different biomechanical conditions.The use of wider skis or, in particular, skis with a large waist width, on a hard or frozen surface, could unfavourably bring the knee joint closer to the end of range of motion in transversal and frontal planes as well as may potentially increase the risk of degenerative knee injuries.The overall results of the abduction and internal rotation in respect to turn radii and ground reaction forces indicated that the knee joint movements are likely one of the key points in alpine skiing techniques. However, the skiing equipment used can still significantly influence the movement strategy.

Estimation of alpine skier posture using machine learning techniques.

High precision Global Navigation Satellite System (GNSS) measurements are becoming more and more popular in alpine skiing due to the relatively undemanding setup and excellent performance. However, GNSS provides only single-point measurements that are defined with the antenna placed typically behind the skier's neck. A key issue is how to estimate other more relevant parameters of the skier's body, like the center of mass (COM) and ski trajectories. Previously, these parameters were estimated by modeling the skier's body with an inverted-pendulum model that oversimplified the skier's body. In this study, we propose two machine learning methods that overcome this shortcoming and estimate COM and skis trajectories based on a more faithful approximation of the skier's body with nine degrees-of-freedom. The first method utilizes a well-established approach of artificial neural networks, while the second method is based on a state-of-the-art statistical generalization method. Both methods were evaluated using the reference measurements obtained on a typical giant slalom course and compared with the inverted-pendulum method. Our results outperform the results of commonly used inverted-pendulum methods and demonstrate the applicability of machine learning techniques in biomechanical measurements of alpine skiing.