PFLEGING, Johannes | On the Foundations of Robotic Use-wear Analysis: Improving Archaeological Methods for Artefact Characterization with Robotics Technology | ETH Zurich, Switzerland

Pages: 148    
Publisher: ETH Zurich (Zurich, 2019)    
Language: English                                                                                                  Methodologies: Archaeology, Robotics, Use-wear Analysis                                              Periods: Lower Palaeolithic, Middle Palaeolithic                                                                      Related Topics: Manipulation Tasks, Methodology, Motion Capture, Hide scraping, Spear thrusting                                                                                                                                        Link:                                  Abstract: In archaeological use-wear analysis, the comparative study of stone tool artefacts and replicas from experiments provides evidence for their function. Thereby it helps to discover the behaviour and evolution of our prehistoric ancestors. However, major advancements in the methodology have been missing since decades. For example, identification of basic kinetic and dynamic parameters of the task motion such as force applied or duration of use is poorly reproducible; a satisfying characterization of the material treated by the tool is not yet possible. We propose a couple of methodological improvements with quantitative evaluation of the variables in use-wear analysis to overcome these limitations. At the centre of our revised methodology stands a robot arm serving as platform for innovative micro-wear experiments. The studies performed with our set-up provide advantages over those involving human subjects, as precise control of the task motion, reproducibility of the experimental conditions and ease of use appear superior. We combine our platform with focus-variation microscopy for topographic surface analysis of experimental replicas. Results with this approach demonstrate the advantage of highly controlled data for micro-wear analysis. We present the influence of force and duration on surface roughness, identifying duration as the dominant factor. To complement the characterization of experimental replicas, we developed a high resolution force sensor array to estimate the load distribution on the replica. The proposed sensor consists of a micro structured silicone membrane, which alters its optical properties when deforming under load. Presuming proper calibration, the force distribution can be inferred from optical photographs of the deformed membrane. We manufactured a sensor prototype for prove of concept studies. Next, we performed quantitative analysis of the human task motion by means of a motion capture system. The recorded trajectories serve as reference for the motion in our robot set-up. In addition, we show that the collected data can be used to deduce biomechanical implications for the user of the tool. In a first study, we compared the different motions between an expert and a non-expert during hide scraping. The results of a second study allowed the reconstruction of Neanderthals hunting behaviour. Wound marks of a cervid bone, dating to 120000 years before present, were analysed and compared with fracture patterns on reference bone targets after impacts with a thrusting spear. We correlated the observed features with the spear kinematics. Finally, we present a framework to generate complex motions for our robot set-up. Tasks relevant for use-wear analysis are characterized by contact switches and frictional contacts. Our approach aims for trajectory optimization with a soft contact model and regularized Coulomb friction to mimic contact interactions of the robot arm with the ground.