Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
tutorials:demo:fetch_and_place [2019/10/18 11:01]
arthur
tutorials:demo:fetch_and_place [2020/06/26 10:52] (current)
gkazhoya Added motivation section
Line 4: Line 4:
  
  
 +===== Motivation =====
  
 +The aim of the tutorial is to give an intuition about the complexity of robot mobile manipulation tasks and about the knowledge that is required to execute actions successfully.
  
 +The tutorial:
 +  - Gives an intuition of what knowledge does the robot need (and how much knowledge) to execute even a simple fetch and place, e.g., robot needs to answer questions such as "where should I stand?",​ "which grasp pose should I use?", etc.
 +  - Teaches how to call existing actions from the CRAM framework, i.e. how to use designators (but not program new ones).
 +  - Shows how many different things can go wrong, and teaches writing simple failure handling strategies.
  
 ===== Setting Up ===== ===== Setting Up =====
Line 1171: Line 1177:
                            ​(object ?​perceived-bottle)))))                            ​(object ?​perceived-bottle)))))
 </​code>​ </​code>​
-You will see that PR2 will successfully grasp the bottle from the grasp pose that we defined.+You will see that PR2 will successfully grasp the bottle from the grasp pose that we defined. Another useful documentation on this point is [[http://​cram-system.org/​doc/​package/​cram_designators]]. Here you can understand further what the '​desig:​an'​ does and why we use it. Designators are a very useful thing in cram, furthermore they are used in every high-level plan
  
 {{:​tutorials:​demo:​diagonal_grasp.png?​600|}} {{:​tutorials:​demo:​diagonal_grasp.png?​600|}}
Line 1212: Line 1218:
  
 Arthur got the following feedback from students: 3h isn't enough to complete the tasks, the intro of the tutorial was too long, designators should be explained. Maybe put a link to those topics at appropriate position. Arthur got the following feedback from students: 3h isn't enough to complete the tasks, the intro of the tutorial was too long, designators should be explained. Maybe put a link to those topics at appropriate position.
 +
 +Alina got the following feedback from students:
 +- an introduction to how to navigate and handle Emacs would have been nice at the start
 +- VM is too slow for some. Laptops'​s we can lend with everything pre-installed would be nice (be it VM or native doesn'​t matter)
 +- Too little time. Nobody finished Exercise 1
 +- visualization of poses could be done better (we used Rviz and "point clicked",​ but rotation remains a problem)
 +- Someone also suggested to do the following: Emacs crash course + basic shortcuts introduction,​ LISP introduction (just show some basic functions since they look very different to other languages. Like, variable definitions,​ let, if, mapcar (it's cool)) and then go to CRAM. (Maybe emphasize rather on plans/​actions which look more like "​natural language"?​)
 +
 +Some questions the students had and some answers we weren'​t sure about
 +- What licence is cram uder (bsd?)
 +- Can it be used without ROS? -> no. unless someone writes something for it since it's open source
 +- How does the PR2 know that he dropped something
 +- Would the belief state work with multiple robots at the same time? -> yes?
  
  
 --></​html>​ --></​html>​