1
00:00:00,090 --> 00:00:07,350
In previous lesson, we have seen how we can solve the issue of requiring too much velocity and acceleration

2
00:00:07,350 --> 00:00:10,680
from actuators that can actually saturate them.

3
00:00:11,520 --> 00:00:18,750
We have solved it by using kinematic scaling, but during trajectory planning we have to consider also

4
00:00:18,750 --> 00:00:23,820
the torque limits and not to saturate actuators by exceeding torque limits.

5
00:00:24,420 --> 00:00:29,700
If we exceeded, we can then use dynamic scaling to solve the issue.

6
00:00:30,210 --> 00:00:36,210
While planning trajectory, you have to consider a nonlinear TS and couplings in robot dynamics.

7
00:00:36,720 --> 00:00:44,310
If your trajectory abides by the torque limits of each individual actuator separately, it doesn't mean

8
00:00:44,310 --> 00:00:50,880
that it will do the same when they are considered together because of non minorities and couplings.

9
00:00:51,390 --> 00:00:55,080
OK, let's start with noting down the robot dynamic equation.

10
00:00:55,500 --> 00:00:58,830
We can separate it for each individual joint as this way.

11
00:00:59,340 --> 00:01:05,850
Now, let's assume that we are given trajectory cue to be executed in time interval of zero two T.

12
00:01:06,540 --> 00:01:13,320
Then we can rewrite dynamic equation Abbu as this way by plugging in the trajectory.

13
00:01:13,950 --> 00:01:23,430
Let's call this part of the equation as tall as I t, because this part depends on not only by configuration,

14
00:01:23,580 --> 00:01:25,770
but also velocity and acceleration.

15
00:01:26,220 --> 00:01:28,980
However, granted, the term depends on the the configuration.

16
00:01:29,880 --> 00:01:37,530
This is the new trajectory v where it is parameterized by function of Sigma T Dash.

17
00:01:37,980 --> 00:01:47,460
As you can see, Sigma Zero, Sigma zero zero and Sigma T Dash is T limbless sigma is just the scaling

18
00:01:47,460 --> 00:01:48,570
function of time.

19
00:01:48,960 --> 00:01:55,680
We will see what's useful for us if we plug this trajectory into the dynamic model of the robot manipulator.

20
00:01:55,680 --> 00:02:03,030
We will get this equation as Q Tilde T Dash is nothing but Q Sigma T Dash.

21
00:02:03,240 --> 00:02:09,600
We can find its derivatives very easily by using chain formula we have seen before.

22
00:02:10,080 --> 00:02:12,990
Here is how we find velocity name live first.

23
00:02:13,200 --> 00:02:19,740
The first derivative of QTL, not T Dash as Sigma T Dash is nothing but t.

24
00:02:19,920 --> 00:02:25,240
We can write the derivative of Q with respect to sigma as Q T.

25
00:02:25,770 --> 00:02:29,790
Now let's find the second derivative of the QTL that he dash.

26
00:02:31,050 --> 00:02:36,570
It is easy to find that you have to just take another derivative of Q Tilde, the dhoti dash.

27
00:02:37,020 --> 00:02:40,440
As you can see, we have utilized the chain rule again.

28
00:02:41,010 --> 00:02:46,470
I specified corresponding terms with colors so you can easily understand what's happening.

29
00:02:46,810 --> 00:02:49,770
No, the equation can be rewritten in this way.

30
00:02:50,040 --> 00:02:56,580
As we have formed velocity and acceleration, we can plug them into the equation 1.0 and obtain this

31
00:02:56,580 --> 00:02:57,540
new equation.

32
00:02:57,960 --> 00:02:59,490
Don't let it scare you.

33
00:02:59,700 --> 00:03:04,650
It's very easy that you can stop video and try to follow the each step.

34
00:03:05,400 --> 00:03:09,000
You can easily notice, though, as I term here again.

35
00:03:09,540 --> 00:03:13,410
This will help us to ride the equation in a much simpler way.

36
00:03:13,860 --> 00:03:14,640
And not this.

37
00:03:14,640 --> 00:03:21,300
Also that gravity will not be affected by time scaling as it depends on the configuration.

38
00:03:21,660 --> 00:03:28,760
Indeed, in previous slide, you have seen that Sigma only appears in and after the first derivatives

39
00:03:28,770 --> 00:03:35,610
of the trajectory, so we can write equation in a much simpler way, as in this equation 1.1.

40
00:03:36,060 --> 00:03:42,600
As you can see, we did not consider gravity torque because it is not affected by the scaling eval appended

41
00:03:42,600 --> 00:03:43,350
at the end.

42
00:03:43,860 --> 00:03:47,100
Now it's time to use scaling function of sigma.

43
00:03:48,090 --> 00:03:50,490
Let's determine it as a linear function.

44
00:03:51,180 --> 00:03:57,570
Keep this in mind you don't have to choose this function as this only you can choose it as your desire.

45
00:03:58,110 --> 00:04:06,240
OK, now let's fund derivatives of the sigma so we can plug them in Eq. 1.1 and obtain this very simple

46
00:04:06,240 --> 00:04:06,900
equation.

47
00:04:07,320 --> 00:04:10,500
Now you can easily see how we scale the trajectory.

48
00:04:11,010 --> 00:04:13,440
Now you can also include gravity terms.

49
00:04:13,440 --> 00:04:20,910
Also, we can choose scaling coefficients of lambda in two ways either it will be less than one.

50
00:04:21,180 --> 00:04:29,170
In this case, timespan of the trajectory will increase because T equals the number T dash and required

51
00:04:29,170 --> 00:04:31,710
talks to execute the trajectory will decrease.

52
00:04:32,430 --> 00:04:34,730
Otherwise opposite is happening.

53
00:04:34,740 --> 00:04:42,900
Time duration increases and we have to apply more talks to find the trajectory at the right time to

54
00:04:43,010 --> 00:04:43,920
excuse me.

55
00:04:45,270 --> 00:04:49,860
We have to apply more torque in order to execute the trajectory through time.

56
00:04:50,970 --> 00:04:56,700
OK, OK, I know you want to see it in an example, so let's see an example.

57
00:04:58,410 --> 00:04:59,610
So assume that we.

58
00:04:59,720 --> 00:05:07,020
Are given the trajectory to be executed by third degree of freedom, robot arm in two time units with

59
00:05:07,020 --> 00:05:11,420
torque limits of tall, one bar and tall to bar for each one.

60
00:05:11,790 --> 00:05:16,920
However, after we have calculated the trajectory, we see that the maximum torque while is required

61
00:05:16,920 --> 00:05:22,110
to execute the trajectory from the joint actuators are higher than the thresholds.

62
00:05:22,500 --> 00:05:23,670
What we have to do?

63
00:05:24,240 --> 00:05:26,910
Surely we have to dynamically scale trajectory.

64
00:05:27,480 --> 00:05:28,650
This is how we do it.

65
00:05:29,190 --> 00:05:32,040
First, let me not bomb the equation here again.

66
00:05:32,580 --> 00:05:38,610
From this equation, it's obvious that we can find lambda by just dividing torque limits with maximum

67
00:05:38,610 --> 00:05:43,290
torque values because we want to decrease talks to the threshold values.

68
00:05:43,710 --> 00:05:49,980
And surely we have to find the minimum, not maximum, because if we find the maximum, then one of

69
00:05:49,980 --> 00:05:52,920
the joints will saturate because of high torque value.

70
00:05:53,430 --> 00:06:02,250
After we have formed the lambda, we can change time to T Dash by scaling the original time span.

71
00:06:04,230 --> 00:06:08,490
You can also notice that we didn't have to plan the trajectory again.

72
00:06:08,880 --> 00:06:11,460
We just scaled a time span, and that's it.

73
00:06:11,850 --> 00:06:15,660
We are in the limits of the torch by dynamic scaling.

74
00:06:15,690 --> 00:06:21,990
You can also calculate optimal trajectory in terms of execution time, namely, if required, talks

75
00:06:21,990 --> 00:06:23,880
are lower than the thresholds.

76
00:06:24,120 --> 00:06:31,080
This means that you don't use full power of your actuators so you can scale down the time span and get

77
00:06:31,080 --> 00:06:34,860
faster trajectory execution without surpassing the limits.

78
00:06:35,460 --> 00:06:41,820
You can investigate this problem and more on trajectory planning from this book that I have used for

79
00:06:41,820 --> 00:06:42,390
this course.

80
00:06:42,690 --> 00:06:45,000
It explains everything in a simple manner.

81
00:06:45,360 --> 00:06:49,770
So for more examples, you can refer this book for the problem.

82
00:06:49,770 --> 00:06:54,000
I have set a minute before I go to the page 240.

83
00:06:54,010 --> 00:06:56,250
An example 5.6.
