Below are the instructions on how to run the simulated symbolic reasoner (to test probabilistic planning under noisy conditions and with different heuristics).

• compile yarp with Python 3 support (see http://www.yarp.it/yarp_swig.html and robotology/yarp#1149)

• configure the poeticon repository (https://github.com/robotology/poeticon) with the CMake option BUILD_simulation set to ON. Compile and make install

simple3 scenario	complex6 scenario

• choose an initial visual scenario, e.g., simple3 or complex6, let's call it $SCENARIO

• choose a planner heuristic configuration among:

  • no heuristics
  • creativity
  • adaptability
  • creativity and adaptability

The above choices determine which application XML from poeticon-simulation/app/scripts will be used, for example: sim_complex6_creativity+adaptability.xml

• choose the types and amounts of simulated noise and set them in the poeticon context directory (obtained by: yarp-config context --where poeticon). In there, create a symbolic link with the name dummy_activity_$SCENARIO.ini (e.g., dummy_activity_complex6.ini), which should link to one of the existing noise definitions. For example: ln -s dummy_activity_complex6_onearm_050.ini dummy_activity_complex6.ini

Note: the numbers at the end of the filename, and inside it, refer to the reliability, which is the opposite of noise, i.e., (1 - noise)

Note: the name of the link must correspond to the --from argument as defined in the XML next to the dummyActivityInterface module

• in a terminal, start yarpserver

• in a terminal, start yarprun --server /enigma (this is the name of the yarprun server used during experiments; if you want to change it, change the variable yarprun_port in the Python script as well)

• in a terminal, start yarpdataplayer, open the chosen scenario (e.g., simple3 or complex6), select Options->Repeat, then reproduce

• in a terminal, call the Python script simpoeticon.py with the number of desired experimental episodes next to the -n option, for example: ./simpoeticon.py -n 50

• to see other options, call ./simpoeticon.py --help

• the output of each experimental episode will be printed with the sequence of motor actions and the corresponding metrics data structure ex(#good,#total[,success]), where #good is the number of successful motor actions, #total is the number of total attempted motor actions including failed ones, success is a boolean indicating whether the system attained the goal autonomously or not (this boolean is only printed when its value is fail i.e. false). Example output:

Starting experiment 1/20
[...]
Result of experiment 1/20: ex(4,6)
Full log:
take Ham left=SUCCESS
put Ham Bun-bottom=SUCCESS
take Bun-top left=FAIL
take Bun-top left=FAIL
take Bun-top left=SUCCESS
put Bun-top Ham=SUCCESS
Result: ex(4,6)
Starting experiment 2/20
Result of experiment 2/20: ex(4,4)
Full log:
take Ham left=SUCCESS
put Ham Bun-bottom=SUCCESS
take Bun-top left=SUCCESS
put Bun-top Ham=SUCCESS
[...]

• the sequence of ex() metrics data structures is written to file and can be used to extract statistics - see example visualization below

• each colored line corresponds to a planner heuristic configuration: magenta = no heuristics, blue = creativity, red = adaptability, green = creativity and adaptability. Each line contains five markers corresponding to different levels of noise. Contact me for more information

Copyright: (C) 2012-2015 POETICON++, European Commission FP7 project ICT-288382
Copyright: (C) 2018 VisLab, Institute for Systems and Robotics, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal
Author: Giovanni Saponaro [email protected]
CopyPolicy: Released under the terms of the GNU GPL v2.0 or later.