Till KTH:s startsida Till KTH:s startsida

Milestone specifications

Milestones

Milestones have been defined to help you meet the requirements of the project on time (we call them milestones because we seem them as your milestones rather than our tollgates). You do not have to meet the milestones before the deadlines but years of experience shows that it helps the groups a lot.

The tasks corresponding to the milestones may seem simple but, do not underestimate them and use your time wisely. Do not leave the things for the final week, this is a real situation where your intelligence is not the only thing that matters but your ability to organize your work and to account for the unpredictable.

I have put the dates for the milestones into this document to convenience. In case of differences, the ones in BILDA are the real ones. Please report any difference so that I can fix it.

For all milestones you should be ready for inspection and be able to back up what you claim to have accomplished.

Milestone 0 (M0) : Oct 23

Using the RGB-D sensor you should be able to follow a hand held infront of the robot (or something else that you can move close in front of the robot. THINK the closest object infront of the sensor). The robot should always face the hand and move towards it until the distance is less than some distance of your choice at which point it should stop (still adjusting angle if needed). You can think of the hand as a moving "goal point". You do not need to recognize a hand. Think of it as the closest object at sensor height so it could be something else like a table leg. The robot should be in one piece but otherwise no special requirements are placed on it.

Deliverable: Video uploaded to BILDA and youtube

 

Milestone 1 (M1) : Nov 12

You should

  • have more or less complete mechanical structure (remember that we do not want you to focus on the mechanics).
  • be able to do wall following in the maze (ie not just follow a straight wall)
  • detect and identify at least one object

 

Deliverable: Video uploaded to BILDA and youtube

 

Milestone 2 (M2) : Nov 26

You should be able to build a map of the environment and mark the position of detected and identified objects the map. You need to come up with some way to show that you have a map and that you have the positions and id of objects. Record an 'evidence' rosbag with camera images of the detected/identified objects following the specification in this wiki page. Your robot should speak every time it detects/identifies a new object.

Deliverables:  Video AND rosbag uploaded to BILDA and/or youtube

 

Milestone 3 (M3) : Dec 05

Show that you can localize in the maze given a map constructed in M2 and find a good path to at least one object that was mapped before. It is important that it is visible that you can localize in your map.

Suggestion for the video: The idea with this milestone is that you should have a robot that can use the map to navigate around in the maze. Explore the maze (likely needs to be speeded up), go back to the start and go to a non-trivial object (ie not next to the start and, not using right/left-wall-following, etc).

Record an 'evidence' rosbag with camera images of the detected/identified objects following the specification in this wiki page. Your robot should speak every time it detects/identifies a new object.

Deliverables:  Video AND rosbag uploaded to BILDA and/or youtube

Teacher Patric Jensfelt created page 23 September 2014

Patric Jensfelt edited 14 October 2014

Milestones

Milestones have been defined to help you meet the requirements of the project on time (we call them milestones because we seem them as your milestones rather than our tollgates). You do not have to meet the milestones before the deadlines but years of experience shows that it helps the groups a lot.

The tasks corresponding to the milestones may seem simple but, do not underestimate them and use your time wisely. Do not leave the things for the final week, this is a real situation where your intelligence is not the only thing that matters but your ability to organize your work and to account for the unpredictable.

For all milestones you should be ready for inspection and be able to back up what you claim to have accomplished.

Milestone 0 (M0)

Using the RGB-D sensor you should be able to follow a hand held infront of the robot (or something else that you can move close in front of the robot. THINK the closest object infront of the sensor). The robot should always face the hand and move towards it until the distance is less than some distance of your choice at which point it should stop (still adjusting angle if needed). You can think of the hand as a moving "goal point". You do not need to recognize a hand. Think of it as the closest object at sensor height so it could be something else like a table leg. The robot should be in one piece but otherwise no special requirements are placed on it.

Deliverable: Video uploaded to BILDA and youtube

 

Milestone 1 (M1)

You should


* have more or less complete mechanical structure (remember that we do not want you to focus on the mechanics).
* be able to do wall following in the maze (ie not just follow a straight wall)
* detect and identify at least one object
 

Deliverable: Video uploaded to BILDA and youtube

 

Milestone 2 (M2)

You should be able to build a map of the environment and mark the position of detected and identified objects the map. You need to come up with some way to show that you have a map and that you have the positions and id of objects.

Deliverables:  Video uploaded to BILDA and youtube

 

Milestone 3 (M3)

Show that you can localize in the maze given a map constructed in M2 and find a good path to at least one object that was mapped before. It is important that it is visible that you can localize in your map.

Suggestion for the video: The idea with this milestone is that you should have a robot that can use the map to navigate around in the maze. Explore the maze (likely needs to be speeded up), go back to the start and go to a non-trivial object (ie not next to the start and, not using right/left-wall-following, etc).

Deliverables:  Video uploaded to BILDA and youtube