Project

General

Profile

0315 » History » Version 2

Pierre Narvor, 2018-03-16 10:40

1 1 Pierre Narvor
h1. 03/14-15 : PoM/Prism Discussion (details)
2
3
*Participants :* 
4
Simon, Ellon, Quentin, Pierre
5
6 2 Pierre Narvor
*Goal of the meeting :* Discuss the PoM/Prism Details. 
7 1 Pierre Narvor
8 2 Pierre Narvor
*Note :* 
9
After the first meeting, Prism design appeared more complicated than previously thought. Quentin and Pierre met afterward to think about a potential solution.
10
Pose and transform are used interchangeably in the following and are always accompanied with a parent frame id, a child frame id and an uncertainty matrix.
11 1 Pierre Narvor
12
 
13
 
14
15 2 Pierre Narvor
h2. 1. Summary of the objectives and constraints (without design details)
16 1 Pierre Narvor
17
 
18
19 2 Pierre Narvor
*1. Nodes producing raw data*
20 1 Pierre Narvor
21 2 Pierre Narvor
* Joints : produce poses between joint base and joint head.
22
* Sensors : produce sensor data (point clouds, etc...) .
23 1 Pierre Narvor
24
 
25
26 2 Pierre Narvor
*2. Nodes receiving processed data*
27 1 Pierre Narvor
28 2 Pierre Narvor
* Localization DFPCs, receives stamped data : raw sensor data stamped with a time stamp and a pose(RobotBase->Sensor)
29
* MightlyLocalizer (part of POM) receives time stamp : time stamp at sensor data acquisition date.
30 1 Pierre Narvor
31
 
32
33 2 Pierre Narvor
*3. Constraints*
34 1 Pierre Narvor
35 2 Pierre Narvor
* Centralized time : POM is the master of time. Every time stamp has to come from POM.
36
* Publish/Subscribe only : No direct services between nodes allowed.
37
* Sensors only know their frame id and are not managing transform propagation between sensor frame and robot frame. These are managed elsewhere (prism).
38 1 Pierre Narvor
39
40 2 Pierre Narvor