Examples for Reading Reports

Here are a few paper summaries that I have written this semester. Note that these (with the exception of the first) are just summaries - I structure my own "critiques" in terms of issues for possible discussion in class.

Starting with the next reading assignment, I will select some students - written reading reports to put on the web as examples.

Bohringer, Bhatt, and Goldberg, ``Sensorless ManipulationUsing Transverse Vibrations of a Plate,'' ICRA 95.

Summary

This paper describes a novel method for feeding planar parts based on the fact that parts resting on a vibrating plate will align to a node of the vibration. The vibrating plate is treated as producing a force field which acts on the part; this provides a model which can be used for simulation and analysis of part behavior. For simple force fields, the authors describe conditions for determining the possible rest configurations of a part. For the more general force fields that arise in practice, they use numerical simulation to determine rest configurations. Based on Goldberg's sensorless grasp plans and the actuator array plans by B\"ohringer, et al.\ and Will and Liu, they suggest that sensorless plans can be automatically generated that bring a part from any initial configuration to a single goal configuration.

They have done an experimental implementation in which the nodes of the plate for different vibration frequencies are experimentally measured, and the force field is assumed to grow linearly with distance from the node. They report that their simulations with this force field closely match experimental results, and demonstrate an alignment plan which takes a triangular part from two different initial orientations to a single goal configuration.

Critique

This paper is a good example of developing a model for a physical process and making use of that model to accomplish a task. Although there are many elements lacking from an ideal and complete theory that maps directly into a practical experimental implementation (an admittedly lofty goal), they present the basic analysis and lay out directions for further work. Their problem is well motivated and shows promise of developing useful results.

Sakaguchi, Masutani, and Miyazaki, ``A Study on Juggling Task,'' IROS 91.

This paper describes analysis, simulations, and an experimental robotic implementation of juggling. The task is to juggle multiple balls using a single hand. The authors began by recording two human subjects juggling two balls; they report several observations including the fact that their hand trajectories approximate an ellipse. The problem is treated in the plane, and they begin by combining the equations for an elliptical hand trajectory with those for a ball thrown by the hand. The resulting equations relate the throwing and catching phase (in the elliptical trajectory), the throwing velocity, and the throwing acceleration to geometric parameters of the ellipse.

From their observations of human juggling, they formulate some conditions on Theta_t, the phase trajectory of the hand: that it is periodic, is increasing, and is in C^3. They pick two functions that satisfy these conditions and perform some simulations, varying various parameters: the number of balls, the magnitude of the velocity oscillation, and the throwing phase.

They have implemented this system using a parallelogram manipulator with a cone shaped cup for the hand. It appears that the actual implementation used some learning method to refine or generate the hand trajectory.

Buhler and Koditschek, ``From Stable to Chaotic Juggling:Theory, Simulation, and Experiments''

This paper studies a system that ``juggles'' a puck on an inclined plane using a rotary bat. One goal of the paper is to show the stability of the ``mirror law'' used to control the bat which nonlinealy reflects the position of the puck.

The authors develop the open loop model of a one dimensional model of the system and then form the closed loop model with the mirror law. The friction between the puck and the plane is included in this model. They then call upon some advanced results from chaos theory in order to show a condition for global stability of the system. This condition essentially keeps the system below the level of the first bifurcation in juggling behavior.

They show experimental results in which the actual parameter values at which bifurcations occur match their analytical predictions closely. They have also demonstrated planar juggling using a more generalized version of the mirror law.