Info for spaceTimeTraj.py
-------------------------
This program combines the ability to generate spatial features of a path with
the ability to generate temporal features of a path.
python prog.py goalx goaly goalangle goaldist numsines amp exp gaze fraction startvel acc [hes_dur]
Parameters
----------
GOAL SETTINGS:
1. Goal x
- x location the goal that the CoBot's path approaches
2. Goal y
- y location the goal that the CoBot's path approaches
3. Goal angle
- CoBot's final orientation, relative to the goal. If you don't want the final
orientation to differ from the path orientation, enter None instead of a float.
4. End pose distance to goal
- CoBot's distance from the goal at its final position (stopped at
end of path). Note that this distance is subtracted along the central
axis of the path rather than along the path itself.
SPACE FEATURES:
5. NumSines
- number of periods of the sine curve that occur over the course of the path
6. Amplitude
- peak distance of the path from the center axis
7. Exponent
- you should set this to 1 unless aiming for a 'hooked' path that moves
gradually away from the center axis before turning back towards the goal
near the end. The higher the exponent, the sharper the turn.
8. Gaze
options:
- path: CoBot orients along the center of the path
- goal: CoBot always orients toward goal as it travels along path
- other possibly useful orientations can be found in trajectoryWithSpace.py
TIME FEATURES:
9A. Fraction of path (components should sum to 1, see below)
- amount of path this segment will be (half of the path is .5)
- note: this number on all segments must add to 1 (if the first segment
has a decimal amount of .3 and there are two segments, the second must
have a decimal amount of .7)
9B. Start velocity
- the velocity at which CoBot should travel at the start of this segment
9C. Acceleration
- acceleration along this path segment
Repeat params 9A-C for as many path segments as you like.
optional final argument - if a stopping hesitation is created (a segment with
arguments 0 0 0) a final argument should be added which is the duration of
the hesitation in seconds (though in execution this number of seconds
is not completely accurate). If this final argument is not included,
CoBot will very briefly stop at the hesitation.
Notes
-----
- The program can currently handle only one stopping hesitation on a single
path.
- The duration of the stopping hesitation is not totally accurate and may
vary between the simulator and the real robot.
- The program will not run if the decimal amounts of all segments do not
add up to 1.
- CoBot's minimum speed is 0.2 m/s, if it tries to decelerate below that speed
an error message will say "Reached minimum speed, continuing at 0.2m/s"
- CoBot does have a maximum speed, and a similarly if it tries to accelerate
above that speed it will continue at the max speed (although I don't know
what that speed is)
Examples
--------
-Endpose features: ends with an orientation angled along the path and stops 0.5 meters from the goal
-Spatial features: straight linear path
-Temporal features: constant velocity of 1m/s with a 1.5 second stopping hesitation halfway along path
python spaceTimeTraj.py x y None 0.5 0.0 0 1 path .5 1 0 0 0 0 .5 1 0 1.5
- Endpose features: ends with an orientation turned -1.5 radians away from the goal, stops 0 meters from the goal
- Spatial features: single sine wave with an amplitude of 0.25 meters, looking along the path
- Temporal features: travels at 1m/s for 60% of the path, then slows to minimum speed for 10% of the path and accelerates along the remaning 30% of path
python spaceTimeTraj.py x y -1.5 0 2.0 .25 1 center .6 1 0 .1 .2 0 .3 .2 1
Ravenna & Laura RISS CoBots
Friday, August 14, 2015
Wednesday, August 12, 2015
Path Title Decoding Chart
| 1st Segment | 2nd Segment | Orientation |
| B (Full amp sine) | C (Full amp sine) | P (Path-oriented) |
| S (Half amp sine) | T (Half amp sine) | G (Goal-oriented) |
| L (Line) | M (Line) |
Hesitation Video Naming Code
| Object | Hesitation | Duration | Distance | Motion |
| R (robot) | W (slowing) | B (brief) | S (social) | Y (yes, rotating) |
| H (human) | T (stopping | L (long) | P (public) | N (no, stationary) |
Tuesday, August 11, 2015
Space Research Goals
Legibility:
How does path shape affect the perceived directness of movement
How does amplitude variance affect the perceived directness of movement
How does orientation affect the perceived directness of movement
Which of these is the most significant with respect to influencing people’s assessment of the path
Storytelling:
Understanding which compound gestures appear most humanlike to a human observer
Understanding which characteristics of paths towards the goal result in which perceived attitudes
Determining what rationale a human gives for CoBot's actions
Attentional Attitudes:
Is there a correlation between the directness of an approach path and the type of perceived attitude
Understand how does the order of path segments affect the perceived attitudes of CoBot
Determining if a change between path segments correlates to a change in attitude, or if an entire compound gesture suggest a single attitude
Possible attitudes towards the goal:
Positive vs Negative
Reluctant vs Enthusiastic
Curious vs Indifferent
Certain vs Uncertain
Focused vs Unfocused
Evasive vs Aggressive
Possible hypotheses:
Participants will associate the switch between the two path segments with a change in attitude.
Participants will rate the lesser amplitude as more direct than the greater amplitude.
Orientation will be a greater influence than path shape for the participants.
Mapping Path Directness to Perceived Attentional Attitudes
Humans observe other people’s body language in order to get an understanding of their thoughts, feelings, and intentions without having to communicate directly with them. This makes human interaction more comfortable and cooperative. While efforts have been made to make human-robot interaction easier by copying human movement to humanoid robots, not all robots are humanoid and have the ability to replicate gestures specific to humanlike body parts. This research is therefore isolating simple x-y-theta movement for non-anthropomorphic robots in order to identify aspects of motion that are inherently expressive.
We’re using concepts of motion from theatre to evaluate robot movement into identifiable characteristics, starting with the Indirect-Direct scale of movement in the domain of Space, with the goal of being able to map these characteristics in a path to specific types of attentional attitudes. To determine these characteristics we generated varying approach paths to an object of interest, according to our deconstruction of a path into several combinable elements, such as shape and orientation. A preliminary pilot was conducted to assess the directness of the different path characteristics, in which observers watched the robot execute different paths towards an object. Indirect ratings were highest for sine paths with path-orientations, and direct ratings were highest for linear paths and goal-orientations. Participants described the robot’s attitudes in similar ways for specific types of directness, and were more likely to engage in storytelling when describing the robot's inner state if the path was more indirect.
Our pending study focuses on compound combinations of path segments to explore how successive contrasting paths might change how the robot’s attitudes are perceived. Overall, this research demonstrates the feasibility of using simple translations and rotations in space to convey humanlike attitudes in non-anthropomorphic robots with limited degrees of freedom.
Friday, August 7, 2015
Research Goals (hesitations study)
Research goals:
Legibility of Laban Efforts:
Understand when people find hesitations to be sudden or sustained.
Understand how sudden or sustained each type of hesitation is.
Understand when people find hesitations to be strong or light.
Understand how strong or light each type of hesitation is.
Understand how the hesitation duration affects how sudden or sustained the hesitation is perceived to be.
Understand how the hesitation duration affects how strong or light the hesitation is perceived to be.
Understand how the hesitation type affects how sudden or sustained the hesitation is perceived to be.
Understand how the hesitation type affects how sudden or sustained the hesitation is perceived to be.
Storytelling:
Understand how type of hesitation affects attributions of surprise.
Understand how type of hesitation affects attributions of excitement.
Understand how type of hesitation affects attributions of certainty.
Understand how type of hesitation affects attributions of timidness.
Understand how type of hesitation affects attributions of tentativeness.
Understand how duration of hesitation affects attributions of surprise.
Understand how duration of hesitation affects attributions of excitement.
Understand how duration of hesitation affects attributions of certainty.
Understand how duration of hesitation affects attributions of timidness.
Understand how duration of hesitation affects attributions of tentativeness.
Understand how proxemic zone affects attributions of surprise.
Understand how proxemic zone affects attributions of excitement.
Understand how proxemic zone affects attributions of certainty.
Understand how proxemic zone affects attributions of timidness.
Understand how proxemic zone affects attributions of tentativeness.
Understand how rotating or stationary relational object affects attributions of surprise, excitement, certainty, timidness, tentativeness.
Understand how duration of hesitation interacts with type of hesitation when participants attribute attitudes to the robot’s motion.
Attitude toward attentional goal:
Understand interaction between proxemic zone and rotation/not of relational object on people’s association of the hesitation as a reaction to the other object.
Robot hesitations express attitudes
People naturally use movement to express their attitudes and internal states and we intuitively read these movements in other people. In robots, expressive motion is important to express humanlike attitudes without human motion. Expressing attitude can help people understand the inner state of the robot - if it’s very busy and in a hurry, concerned about a collision, or lost, etc.
The Laban Effort System is used by actors to reproduce motion, we use it to create gestures for robots. It helps deconstruct movement and we use this to generate gestures for robots that can express humanlike attitudes.
We were interested in time domain features of a robot’s movement along a path. We referred to the Laban Effort system to create velocity settings along a scale from sudden (abrupt) to sustained (gradual) and acceleration settings along a scale from strong (powerful) to light (delicate).
We conducted a pilot study to see how people interpreted various temporal gestures. Our most interesting findings were related to hesitations. Participants observed the robot make slowing and stopping hesitations (the robot either slows to its minimum speed or comes to a full stop) along its path to an attentional goal. People attributed different attitudes to the robot depending on the type of hesitation. Slowing hesitations were associated with uncertainty, tentativeness, and timidity. Full stop hesitations made the robot look more sure of its motion. They were associated with surprise or excitement.
A full study is pending in which we will focus on hesitations, varying hesitation type and other characteristics of the hesitations in order to understand how we can use them to express specific attitudes in robots.
Subscribe to:
Posts (Atom)