-
Notifications
You must be signed in to change notification settings - Fork 36
/
Copy pathISSUE_NOTE.txt
80 lines (48 loc) · 2.76 KB
/
ISSUE_NOTE.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
Feb 2020: Users reporting UR5 broken on master.
>> 3-Feb: solution of UR5 working. just some extra variables in the output: eg th_23
which have solutionmethod = "*none* and no solution equations.
need to prune nodes by nodes.solutionmethod
21-Dec Update: prune by zero length solution method instead!
----------------------------------------
New branch RepairUR5Regression which is newest commit which works with ARM & UR5.
** UR5 doesn't solve trival eqns like th_4 = th_234-th2-th3
-- make sure SOA definition equations are included during solver scanning of eqn lists.
---------------------------------
May 2019: "Arm_3"
* This arm now works ... mostly. The Latex report has a couple of gaps but still produces a workable solution.
* Still 2 open issues with ArmRobo and Lori.
21-Feb-19: Following are open issues:
* simple planar 3-link "Arm_3" fails to properly manage 3-way sum of angles
* python -m ikbtleaves.sum_transform FAILS
* find_xy function (see below) is commented out in ik_classes()(@*)!_#)%%?)
* SofAngles fails some previous test cases in the comprehensive test included in updateL.py
-----------------------------
August 2018 (just prior to releasing the UR5 Fix)
--
Regarding sum of angles changes to handle 3 parallel axis case:
1) there are two very similar codes which are dupes and should be unified:
a) find_xy in ikbtbasics/ik_classes.py *** official version
b) find_xy in sum_transform.py *** DELETED
2) there are two methods that use find_xy to substitue sum-of-angles. One is in sum_transform.py
and another is a method member of the Robot class (!) the later is used in updateL.
We need to eliminate these redundancies so that we can fix the 3-parallel axis case. Strategy:
1) elmininate duplcated codes **Check: 3-Aug-18
2) find a new method other than the set lookup table (or maybe a lookup table with a two list input):
{[th1,th2] : [ th12, th123],\\
[th2,th3] : [th23, th 123, th234] } etc.
BH 30-Jul
New alg:
1) convert two args to strings (i.e. string having the variable NAME) # does this require new class members?
th1 --> 'th1', th23 --> 'th23' etc
2) separate indices (ints) from strings,
th1 --> '1', th23 --> '23'
3) concatenate and sort.
['23', '1'] --> '123
4) search existing unknowns for match to new variable
'th123' ---> < already a defined unknown > or < create new unknown th123>
Second approach ***
1) each unknown has self.n = index of joint. if index > 10, it means a sum of angle:
i.e. if self.n == 234, self is th_234.
2) for two unknowns, a,b:
if sum is a new unknown where sum = th_a + th_b,
sum.n = int(str(a.n)+str(b.n))