Lillian Y. Chang

A kinematic thumb model for the ACT hand

The thumb is essential to the hands function in grasping and manipulating objects. Previous anthropomorphic robot hands have thumbs that are biologically-inspired but kinematically-simplified. In order to study the biomechanics and neuromuscular control of hand function, an anatomical robotic model of the human thumb is constructed for the anatomically-correct testbed (ACT) hand. This paper presents our ACT thumb kinematic model that unifies a number of studies from biomechanical literature. We also validate the functional consistency (i.e. the nonlinear moment arm values) between the cadaveric data and the ACT thumb. This functional consistency preserves the geometric relationship between muscle length and joint angles, which allows robotic actuators to imitate human muscle functionality

Method for Determining Kinematic Parameters of the In Vivo Thumb Carpometacarpal Joint

The mobility of the thumb carpometacarpal (CMC) joint is critical for functional grasping and manipulation tasks. We present an optimization technique for determining from surface marker measurements a subject-specific kinematic model of the in vivo CMC joint that is suitable for measuring mobility. Our anatomy-based cost metric scores a candidate joint model by the plausibility of the corresponding joint angle values and kinematic parameters rather than only the marker trajectory reconstruction error. The proposed method repeatably determines CMC joint models with anatomically-plausible directions for the two dominant rotational axes and a lesser range of motion (RoM) for the third rotational axis. We formulate a low-dimensional parameterization of the optimization domain by first solving for joint axis orientation variables that then constrain the search for the joint axis location variables. Individual CMC joint models were determined for 24 subjects. The directions of the flexion-extension (FE) axis and adduction-abduction (AA) axis deviated on average by 9deg and 22deg, respectively, from the mean axis direction. The average RoM for FE, AA, and pronation-supination (PS) joint angles were 76deg, 43deg, and 23deg for active CMC movement. The mean separation distance between the FE and AA axes was 4.6 mm, and the mean skew angle was 87deg from the positive flexion axis to the positive abduction axis.

Feature selection for grasp recognition from optical markers

Although the human hand is a complex biomechanical system, only a small set of features may be necessary for observation learning of functional grasp classes. We explore how to methodically select a minimal set of hand pose features from optical marker data for grasp recognition. Supervised feature selection is used to determine a reduced feature set of surface marker locations on the hand that is appropriate for grasp classification of individual hand poses. Classifiers trained on the reduced feature set of five markers retain at least 92% of the prediction accuracy of classifiers trained on a full feature set of thirty markers. The reduced model also generalizes better to new subjects. The dramatic reduction of the marker set size and the success of a linear classifier from local marker coordinates recommend optical marker techniques as a practical alternative to data glove methods for observation learning of grasping.

Planning pre-grasp manipulation for transport tasks

Studies of human manipulation strategies suggest that pre-grasp object manipulation, such as rotation or sliding of the object to be grasped, can improve task performance by increasing both the task success rate and the quality of load-supporting postures. In previous demonstrations, pre-grasp object rotation by a robot manipulator was limited to manually-programmed actions. We present a method for automating the planning of pre-grasp rotation for object transport tasks. Our technique optimizes the grasp acquisition point by selecting a target object pose that can be grasped by high-payload manipulator configurations. Careful selection of the transition states leads to successful transport plans for tasks that are otherwise infeasible. In addition, optimization of the grasp acquisition posture also indirectly improves the transport plan quality, as measured by the safety margin of the manipulator payload limits.

Representation of pre-grasp strategies for object manipulation

In this paper, we present a method for representing and re-targeting manipulations for object adjustment before final grasping. Such pre-grasp manipulation actions bring objects into better configurations for grasping through e.g. object rotation or object sliding. For this purpose, we propose a scaling-invariant and rotation-invariant representation of the hand poses, which is then automatically adapted to the target object to perform the selected pre-grasp manipulations. We show that pre-grasp strategies such as sliding manipulations not only enable more robust object grasping, but also significantly increase the success rate for grasping.

Preparatory object rotation as a human-inspired grasping strategy

Humans exhibit a rich set of manipulation strategies that may be desirable to mimic in humanoid robots. This study investigates preparatory object rotation as a manipulation strategy for grasping objects from different presented orientations. First, we examine how humans use preparatory rotation as a grasping strategy for lifting heavy objects with handles. We used motion capture to record human manipulation examples of 10 participants grasping objects under different task constraints. When sliding contact of the object on the surface was permitted, participants used preparatory rotation to first adjust the object handle to a desired orientation before grasping to lift the object from the surface. Analysis of the human examples suggests that humans may use preparatory object rotation in order to reuse a particular type of grasp in a specific capture region or to decrease the joint torques required to maintain the lifting pose. Second, we designed a preparatory rotation strategy for an anthropomorphic robot manipulator as a method of extending the capture region of a specific grasp prototype. The strategy was implemented as a sequence of two open-loop actions mimicking the human motion: a preparatory rotation action followed by a grasping action. The grasping action alone can only successfully lift the object from a 45-degree region of initial orientations (4 of 24 tested conditions). Our empirical evaluation of the robot preparatory rotation shows that even using a simple open-loop rotation action enables the reuse of the grasping action for a 360-degree capture region of initial object orientations (24 of 24 tested conditions).

Selection criteria for preparatory object rotation in manual lifting actions.

ABSTRACT Participants lifted a canister by its handle while balancing a ball on the lid. Experiment 1 allowed object rotation prior to lifting. A lifting comfort zone was measured by the variability in object orientation at lift; its size depended on the object mass and required task precision. The amount of prelift rotation correlated with the resulting change in lifting capability, as measured for different object orientations. Experiment 2 required direct grasping without preparatory rotation. Task completion time and success rate decreased, and initial object orientation affected prelift time. Results suggest that lifting from the comfort zone produces more robust performance at a cost of slower completion; moreover, physical rotation could be replaced by mental planning when direct grasping is enforced.

Templates for pre-grasp sliding interactions

In manipulation tasks that require object acquisition, pre-grasp interaction such as sliding adjusts the object in the environment before grasping. This change in object placement can improve grasping success by making desired grasps reachable. However, the additional sliding action prior to grasping introduces more complexity to the motion planning process, since the hand pose relative to the object does not need to remain fixed during the pre-grasp interaction. Furthermore, anthropomorphic hands in humanoid robots have several degrees of freedom that could be utilized to improve the object interaction beyond a fixed grasp shape. We present a framework for synthesizing pre-grasp interactions for high-dimensional anthropomorphic manipulators. The motion planning is tractable because information from pre-grasp manipulation examples reduces the search space to promising hand poses and shapes. In particular, we show the value of organizing the example data according to object category templates. The template information focuses the search based on the object features, resulting in increased success of adapting a template pose and decreased planning time.

Pre-grasp Interaction for Object Acquisition in Difficult Tasks

In natural manipulation activities of daily living, actions for object grasping must respect several constraints for successful task completion. For example, grasping actions must satisfy at a minimum the reachability of grasp contacts on the object surface, collision avoidance with obstacles, and kinematic as well as strength limits of the hand. In challenging manipulation scenarios with high constraints, direct reaching actions to grasp the object in place may not be sufficient for object acquisition. We have observed that humans use pre-grasp interaction to adjust the object placement during the grasping process. For example, an object may be slid or tumbled on its support surface before the final grasp contacts are achieved. In this chapter we provide an overview of the variety of pre-grasp actions that we have observed from a video survey of human manipulation activities in natural home and occupational environments. We then present our studies of object reorientation by rotation, as a particular type of human pre-grasp interaction. Finally we examine the utility of pre-grasp rotation for increasing object reachability and grasp reuse for a robot manipulator.

Estimating subject-specific parameters for modeling hand joints

In both biomedical and graphics applications, quantifying skeletal motion requires a kinematic model describing the joint location and axis directions. In this talk, we present techniques we have recently developed for determining parameters of three types of anatomic joints. In particular, it is important but difficult to obtain high-quality parameter estimates for the joints of the human hand. Due to the small scale of hand segments, mis-location of joint centers by even 1 cm leads to unacceptable errors in the joint angle measurement. In addition, the axis directions of hand joints are not aligned with standard anatomical planes [1], and each individual has subject-specific axis directions. The errors in axis location and direction from standard estimation techniques result in noticeable differences in reconstructed hand shape and the grasp contact points of the fingers.