Human Affordances of Stacking: Best Placement or Best Outlook? Ameer Ibrahim, Giuseppe Cotugno, Kaspar Althoefer, Thrishantha Nanayakkara ameer.ibrahim@hotmail.co.uk, {giuseppe.cotugno, kaspar.althoefer, thrish.antha}@kcl.ac.uk Centre for Robotics Research, King's College London (London, United Kingdom) Humans face the everyday need of using objects to pursue their goals for simple and common tasks. The possible usages assigned to objects are called affordances (Gibson 1977). (Şahin, Çakmak et al. 2007) proposes a good summary of different classifications and definitions of affordances. (Costantini and Sinigaglia 2011) believe that affordances are driven not only by physical properties of the object but also by its placement and reachability in the space. However, at what extent the geometrical properties of an object can overtake displacement in space? How much influence do geometrical properties have in the process of selecting the objects to complete a task? Do we always try to minimize muscular effort or are we also guided by appearance? Those are the scientific questions that we have addressed in this paper. In this work, an affordance is defined as a possible way to use a reachable object to fulfil a task. We assume that affordances depend mainly on the position of the objects on the table, the availability of objects with similar physical features, their reachability, and their visually perceived physical properties. To find answers to our questions, human subjects were asked to build the tallest stack using a given set of objects. The experiments were executed on 9 right handed adult subjects between the ages of 20 and 30, with no history of any motor impairment diseases. The study was approved by the King's College London Ethical Committee, REC reference Number BDM/12/13-27. Subjects wore the Measurand Inc ShapeHand motion capture glove to collect their movements. Subjects were asked to build the tallest stack out of the given objects, which were laid out on a grid on the table (Figure 1). Five different types of objects were used in the experiment (3 cuboids, 3 cylinders, 2 triangles, 2 pyramids and 1 ball). The objects were laid out on a 3x4. Four different layouts were used as a setting to conduct the investigation (Table 1). 108 trials were executed in total, 12 per subject. The task is to build the stack on the left hand side of the grid in a separate cell. The subjects were given no time limit to complete the task. All the subjects started the taks from the same initial position, according to the calibration requirements of the measurement device, arm and fingers flat in front of their body, thumb alligned to the palm). At the end, subjects return to the same initial position. Subjects were asked to sit upright and move only their dominant arm and hand when stacking. To avoid bending or rotation of the torso, non dominant hand was holding the side of the table. Subjects were asked to hold on to the table with their free hand, to stop them from rotating their torso. Before starting a trial, some time was given to each subject to observe the grid. Initially, there are eleven different possible objects of five different types that can be picked up. Each object has a probability of 1/11 to Figure 1- Example of experimental setup (first layout) and object used in the experiment. Figure 2 - Probability of picking a cuboid as first object given its position on the grid. The higher the probability the more red the cell.