WEBVTT 00:00:00.000 --> 00:00:03.620 So we found two things. The first thing was that 00:00:03.620 --> 00:00:07.940 there was certainly a better device. It was actually 00:00:07.940 --> 00:00:11.259 pretty accurate. The level of accuracy was the 00:00:11.259 --> 00:00:16.219 maximum error that we found was around 0 .01 00:00:16.219 --> 00:00:20.000 meter per second. And we also found that this 00:00:20.000 --> 00:00:22.739 error was unbiased. This means that there was 00:00:22.739 --> 00:00:25.460 no preference in this device to overestimate 00:00:25.460 --> 00:00:28.519 or underestimate the actual velocity of the device. 00:00:29.289 --> 00:00:32.350 during testing while most of the other devices 00:00:32.350 --> 00:00:37.969 that we found actually presented a bias. Hi Francesco, 00:00:38.049 --> 00:00:40.130 it's my pleasure to have you the second time. 00:00:40.210 --> 00:00:42.770 The first time we've been talking muscle activation, 00:00:43.329 --> 00:00:46.450 using straps, not using straps, and this time 00:00:46.450 --> 00:00:49.289 we'll be talking about velocity. But before we 00:00:49.289 --> 00:00:51.289 get there, could you please briefly introduce 00:00:51.289 --> 00:00:53.490 yourself? My name is Francesco Travascio and 00:00:53.490 --> 00:00:56.170 I'm Associate Professor at the Department of 00:00:56.170 --> 00:00:58.189 Mechanical and Aerospace Engineering of the University 00:00:58.189 --> 00:01:01.429 of Miami. I'm also a Master Weightlifter and 00:01:01.429 --> 00:01:04.629 I really enjoy doing studies in weightlifting 00:01:04.629 --> 00:01:07.489 and sports related biomechanics. So we're talking 00:01:07.489 --> 00:01:10.590 velocity, measurement of velocity in weightlifting 00:01:10.590 --> 00:01:14.750 movements. If you would tell me first why you 00:01:14.750 --> 00:01:17.549 decided to set up the study. Well, I was very 00:01:17.549 --> 00:01:21.569 fascinated about the training methodology based 00:01:21.569 --> 00:01:24.930 on barbell velocity. I find it that in principle, 00:01:25.109 --> 00:01:28.250 it seems to me a very rational approach to the 00:01:28.250 --> 00:01:31.230 optimization of the performance of the athlete. 00:01:31.489 --> 00:01:34.209 At that time, there were a number of devices 00:01:34.209 --> 00:01:36.840 that were used for it. tracking the bar per velocity, 00:01:37.120 --> 00:01:39.939 obviously the first thing that we started to 00:01:39.939 --> 00:01:43.299 do was to look into those that could be more 00:01:43.299 --> 00:01:45.840 immediately available for a general user. And 00:01:45.840 --> 00:01:49.299 we looked into camera apps, phones or tablets, 00:01:49.579 --> 00:01:52.239 and we found one that at that time seemed to 00:01:52.239 --> 00:01:56.680 us a very interesting type of app, very, very 00:01:56.680 --> 00:01:59.799 nice. And we also wanted to compare the accuracy 00:01:59.799 --> 00:02:01.799 of this system, obviously with our gold standard 00:02:01.799 --> 00:02:04.879 that is within a gate lab and motion capture. 00:02:05.099 --> 00:02:07.319 system. It was an OptiTrac system that we used. 00:02:07.439 --> 00:02:10.379 Another problem that we found, and we actually 00:02:10.379 --> 00:02:12.900 couldn't find a practical solution for that, 00:02:13.039 --> 00:02:16.139 was the fact that some of these devices are built 00:02:16.139 --> 00:02:20.099 as boxes with a rotational encoder and a wire 00:02:20.099 --> 00:02:22.460 that is attached to the side of the barbell. 00:02:22.539 --> 00:02:25.060 When the barbell is lifted, a cable is pulled 00:02:25.060 --> 00:02:27.680 out of this device, and the rotational encoder 00:02:27.680 --> 00:02:30.360 measures the number of rotations, and there's 00:02:30.360 --> 00:02:32.500 a timer. So by the number of rotations, the timers 00:02:32.500 --> 00:02:34.919 can determine distance and time. to determine 00:02:34.919 --> 00:02:37.120 the velocity. The problem is that when you throw 00:02:37.120 --> 00:02:39.560 the bar on the ground, you have to be very careful 00:02:39.560 --> 00:02:42.639 not to throw it on the device. Eventually, it 00:02:42.639 --> 00:02:45.680 happened. I think Murphy's Law. It happened that 00:02:45.680 --> 00:02:47.800 one of these devices was smashed, and so we had 00:02:47.800 --> 00:02:50.120 to purchase a new one. Together with these devices, 00:02:50.159 --> 00:02:52.719 we also used another very interesting device 00:02:52.719 --> 00:02:56.680 that is a laser positional device. Essentially, 00:02:57.000 --> 00:02:59.039 it's a device that is attached to the end of 00:02:59.039 --> 00:03:03.479 the barbell and has a sensor that emits the light. 00:03:03.530 --> 00:03:06.349 that is reflected by a surface and then receives 00:03:06.349 --> 00:03:08.870 the light back. So by measuring how much time 00:03:08.870 --> 00:03:11.169 it takes from the light to get out of the sensor 00:03:11.169 --> 00:03:13.830 and coming back, you can determine essentially 00:03:13.830 --> 00:03:16.490 the distance of the barber from the floor. It 00:03:16.490 --> 00:03:18.789 was a very neat device. I really liked it. The 00:03:18.789 --> 00:03:21.169 only cumbersome aspect of it is that you really 00:03:21.169 --> 00:03:23.889 need to have a reflecting surface, an appropriate 00:03:23.889 --> 00:03:25.949 reflecting surface. In fact, the device comes 00:03:25.949 --> 00:03:28.349 together with a little mat of a reflecting surface. 00:03:28.689 --> 00:03:31.550 Overall, I think that the iPhone app is probably 00:03:31.550 --> 00:03:34.449 one of those most practical things you can possibly 00:03:34.449 --> 00:03:38.270 do because everyone has a phone nowadays. And, 00:03:38.370 --> 00:03:41.449 you know, it's quite simple to set up. It only 00:03:41.449 --> 00:03:44.889 requires to establish a distance of around 3 00:03:44.889 --> 00:03:46.949 to 10 meters. So we are talking about between 00:03:46.949 --> 00:03:50.229 10 and 30 feet distance from the lifter. Possibly 00:03:50.229 --> 00:03:53.669 being on the side of the lifter, catching the 00:03:53.669 --> 00:03:55.849 thermal weight in the surgical plate. And it's 00:03:55.849 --> 00:03:58.169 quite practical. All right. So let's bring the 00:03:58.169 --> 00:04:01.789 design of the study together. Could you list 00:04:01.789 --> 00:04:04.509 the devices and then what you compared it to 00:04:04.509 --> 00:04:09.360 to get the accuracy? Sure. So we had a total 00:04:09.360 --> 00:04:12.460 of five devices. We had what is called the Jim 00:04:12.460 --> 00:04:16.660 Aware RS, a Vitro, a Gravity Box. And then those 00:04:16.660 --> 00:04:20.740 three essentially are boxes with a cable. The 00:04:20.740 --> 00:04:23.220 principle of function is always the same. Then 00:04:23.220 --> 00:04:26.420 we had the Kinetic Flex, the device that is applied 00:04:26.420 --> 00:04:29.079 on the side of the barbell. And then we had the 00:04:29.079 --> 00:04:32.680 phone app that was called the WL Analysis, Reclinic 00:04:32.680 --> 00:04:36.379 analysis. So three linear transducers, one laser. 00:04:36.459 --> 00:04:40.040 the laser -based measurement tool and one phone 00:04:40.040 --> 00:04:43.120 app, so camera, 2D camera -based measurement. 00:04:43.300 --> 00:04:44.800 Yeah, that is correct. And your gold standard 00:04:44.800 --> 00:04:47.139 was? Our gold standard was our motion capture 00:04:47.139 --> 00:04:51.040 system, an Optidrug system that included 12 cameras. 00:04:51.339 --> 00:04:53.800 Awesome. Obviously, to make it work, track the 00:04:53.800 --> 00:04:56.199 movement of the barbell, we had to apply a marker 00:04:56.199 --> 00:04:58.639 to the barbell. We applied the marker to the 00:04:58.639 --> 00:05:00.939 side of the barbell. It's essentially, it's a 00:05:00.939 --> 00:05:03.339 reflective sphere. We performed the measurement, 00:05:03.480 --> 00:05:06.319 all the measurements simultaneously. We had on 00:05:06.319 --> 00:05:09.240 one side, the two linear transducers and the 00:05:09.240 --> 00:05:11.500 reflective marker. On the other side, one linear 00:05:11.500 --> 00:05:14.540 transducer and the Kinect Flex. Awesome. And 00:05:14.540 --> 00:05:16.500 obviously the phone. Yes, the phone on the side 00:05:16.500 --> 00:05:19.899 and 12 cameras around for the touching system. 00:05:20.079 --> 00:05:22.100 Correct, yes. Now, who were the participants? 00:05:22.319 --> 00:05:25.480 Who did you invite to the study? Those were athletes 00:05:25.480 --> 00:05:28.319 competing at regional and national levels. Most 00:05:28.319 --> 00:05:31.939 of them were from a local gym. But there were 00:05:31.939 --> 00:05:34.800 a wee session. We advertised the study in all. 00:05:34.889 --> 00:05:37.949 Florida in our state and actually we were very 00:05:37.949 --> 00:05:41.430 pleased by a couple of athletes that came from 00:05:41.430 --> 00:05:44.730 Tampa which is located five hours driving from 00:05:44.730 --> 00:05:48.389 Miami and they explicitly came to Miami and they 00:05:48.389 --> 00:05:51.850 drove for five hours just to be part of the study 00:05:51.850 --> 00:05:55.490 and this was really incredible. In terms of comparing 00:05:55.490 --> 00:05:58.490 the devices what measurements did you decide 00:05:58.490 --> 00:06:01.389 to use and what movements did you pass? So for 00:06:01.389 --> 00:06:05.889 the measurements we essentially we the barbell 00:06:05.889 --> 00:06:11.810 velocity in clean and in snatch. We left to the 00:06:11.810 --> 00:06:14.829 participants to decide whether they wanted to 00:06:14.829 --> 00:06:17.790 power or to execute the full movement of all 00:06:17.790 --> 00:06:20.730 the pieces. And we wanted to measure the barbell 00:06:20.730 --> 00:06:24.670 velocity at different percentages of the maximum 00:06:24.670 --> 00:06:27.370 effort just to see if there are any changes in 00:06:27.370 --> 00:06:30.670 the accuracy of the devices as the velocity increases 00:06:30.670 --> 00:06:34.980 or decreases. So we decided to make sure that 00:06:34.980 --> 00:06:39.300 every lift was a proper lift. And also for probably 00:06:39.300 --> 00:06:41.959 safety reasons, we wanted to limit the percentage 00:06:41.959 --> 00:06:45.759 of effort to 70%. So we had lifts executed at 00:06:45.759 --> 00:06:49.680 40 % and at 70%. Essentially at 40%, probably 00:06:49.680 --> 00:06:51.939 the barbell velocity at that point is probably 00:06:51.939 --> 00:06:55.000 quite high, I would say. But at 70 % still, you 00:06:55.000 --> 00:06:57.319 have a pretty high speed of the barbell. The 00:06:57.319 --> 00:07:00.860 numbers, what are the average, more like median 00:07:00.860 --> 00:07:03.740 velocity of the barbell? would be at 40 %? We 00:07:03.740 --> 00:07:06.420 are in the range of two meters per second. There 00:07:06.420 --> 00:07:09.800 are some values of barber velocity that have 00:07:09.800 --> 00:07:12.560 been measured in the past, many, many years ago. 00:07:12.759 --> 00:07:16.060 And well, I mean, but those measurements were 00:07:16.060 --> 00:07:19.220 performed on elite athletes. And obviously, I 00:07:19.220 --> 00:07:23.259 believe that the elite is always, you know, biomechanically 00:07:23.259 --> 00:07:25.980 speaking, is a different animal rather compared 00:07:25.980 --> 00:07:28.399 to the normal population of excellent athletes, 00:07:28.500 --> 00:07:30.360 but still at regional and national levels. They 00:07:30.360 --> 00:07:32.879 are probably more similar. humans. In terms of 00:07:32.879 --> 00:07:35.860 clean, we are talking at 40%, we are talking 00:07:35.860 --> 00:07:39.300 about 2 .3 meters per second. We take it at 70 00:07:39.300 --> 00:07:44.779 % is about 1 .93 meters per second. And for the 00:07:44.779 --> 00:07:47.060 snatch, it's a little bit faster. It's about 00:07:47.060 --> 00:07:53.939 2 .5 or 40 % and around 2 .2, around 70%. Awesome. 00:07:54.079 --> 00:07:56.620 Was it mean velocity or peak velocity you were 00:07:56.620 --> 00:07:58.959 measuring? No, we are talking about peak velocity. 00:07:58.959 --> 00:08:01.370 We were interested in measuring the peak velocity 00:08:01.370 --> 00:08:04.329 because it would just make more sense in my opinion 00:08:04.329 --> 00:08:07.829 within generally in powerlifting the measurements 00:08:07.829 --> 00:08:09.889 that are reported refer to the peak velocity 00:08:09.889 --> 00:08:11.910 as well as the mean velocity within the entire 00:08:11.910 --> 00:08:15.009 lift but we wanted to focus only on a variable 00:08:15.009 --> 00:08:18.509 that in our view made more sense for the sport 00:08:18.509 --> 00:08:21.170 that is essentially the peak velocity which expresses 00:08:21.170 --> 00:08:23.870 the maximum power you can express we measure 00:08:23.870 --> 00:08:28.110 the three repetitions per lift per at 40 percent 00:08:28.110 --> 00:08:30.550 three repetitions per lift that's seven percent 00:08:30.550 --> 00:08:32.809 and we averaged the measurements and essentially 00:08:32.809 --> 00:08:36.870 what we did was to measure these velocities with 00:08:36.870 --> 00:08:39.370 all these devices these peak velocities with 00:08:39.370 --> 00:08:42.029 all these devices and compare them to the gold 00:08:42.029 --> 00:08:44.570 standard that was the motion capture system and 00:08:44.570 --> 00:08:47.350 calculate the error essentially we calculated 00:08:47.350 --> 00:08:50.129 that the error as the difference between the 00:08:50.129 --> 00:08:52.509 measurement with the device versus the measurement 00:08:52.509 --> 00:08:55.610 with the motion capture system and the errors 00:08:55.610 --> 00:08:59.230 made that we found for all the values that we 00:08:59.399 --> 00:09:02.759 All the leads that we measured would vary between, 00:09:03.240 --> 00:09:07.720 in the lowest case, like 0 .01 meter per second 00:09:07.720 --> 00:09:11.340 up to almost 0 .2 meters per second. Let's start 00:09:11.340 --> 00:09:14.399 from the most important thing. So which one, 00:09:14.620 --> 00:09:18.039 which of the devices, if any, was the most accurate? 00:09:18.340 --> 00:09:21.940 So we found two things. The first thing was that 00:09:21.940 --> 00:09:26.059 there was certainly a better device. That was 00:09:26.059 --> 00:09:28.500 the GymAware. It was actually pretty accurate. 00:09:28.620 --> 00:09:32.340 The level of accuracy was the maximum error that 00:09:32.340 --> 00:09:37.259 we found was around 0 .01 meter per second. And 00:09:37.259 --> 00:09:41.100 we also found that this error was unbiased. This 00:09:41.100 --> 00:09:43.879 means that there was no preference in this device 00:09:43.879 --> 00:09:47.440 to overestimate or underestimate the actual velocity 00:09:47.440 --> 00:09:49.779 of the device of the barbell during testing, 00:09:50.000 --> 00:09:52.799 while most of the other devices that we found 00:09:52.799 --> 00:09:56.360 actually presented a bias, meaning that the flex 00:09:56.360 --> 00:10:01.179 device, owner and the truth box presented a negative 00:10:01.179 --> 00:10:03.460 bias meaning that they tended to underestimate 00:10:03.460 --> 00:10:07.019 the actual peak velocity and these underestimations 00:10:07.019 --> 00:10:12.799 were up to 0 .18 0 .17 meter per second now in 00:10:12.799 --> 00:10:16.240 order to get a sense of how important is this 00:10:16.240 --> 00:10:20.960 error if we would extrapolate The value of the 00:10:20.960 --> 00:10:26.580 velocity of an athlete from 60 % to 100 % represents 00:10:26.580 --> 00:10:29.220 the one repetition max. When we are in the region 00:10:29.220 --> 00:10:33.340 of the 95 % to 100%, which is the region that 00:10:33.340 --> 00:10:35.779 most matters to us because it's the peak of the 00:10:35.779 --> 00:10:38.100 performance of this person, that change in the 00:10:38.100 --> 00:10:41.039 power back velocity could be of 0 .1 billion 00:10:41.039 --> 00:10:44.539 per second or probably less. So essentially having 00:10:44.539 --> 00:10:49.740 an error for almost a double, 0 .15, 0 .17. represents 00:10:49.740 --> 00:10:52.519 a great limitation in this device for this particular 00:10:52.519 --> 00:10:55.840 application. Regarding, we actually was, I mean, 00:10:55.860 --> 00:10:58.659 obviously all these devices have a different 00:10:58.659 --> 00:11:01.139 cost associated to it, right? So the GymAware 00:11:01.139 --> 00:11:03.659 device was represented the most expensive type 00:11:03.659 --> 00:11:07.639 of device and something that is probably not 00:11:07.639 --> 00:11:10.679 immediately accessible to the general user because 00:11:10.679 --> 00:11:13.139 we are talking about around $2 ,000 or $3 ,000 00:11:13.139 --> 00:11:16.500 to purchase it versus a phone app that is free 00:11:16.500 --> 00:11:20.470 or $9 without. advertisements right so we were 00:11:20.470 --> 00:11:22.529 a little bit disheartened when we found that 00:11:22.529 --> 00:11:25.429 the phone app actually presented this big limitation 00:11:25.429 --> 00:11:27.970 but i believe that the main limitation of the 00:11:27.970 --> 00:11:31.029 phone app is not on the algorithm that actually 00:11:31.029 --> 00:11:33.610 calculates the velocity of the barbell the limitation 00:11:33.610 --> 00:11:36.950 is essentially in frame per second that the phone 00:11:36.950 --> 00:11:40.100 can can acquire. So, initially, if you could 00:11:40.100 --> 00:11:43.399 have a phone that has, or a tablet, or any device 00:11:43.399 --> 00:11:46.899 that has a larger frame per fps compared to a 00:11:46.899 --> 00:11:50.259 regular iPhone or an Android, it is highly possible 00:11:50.259 --> 00:11:53.559 that you could significantly reduce the error 00:11:53.559 --> 00:11:55.960 in the estimate. The advantage of this, so one 00:11:55.960 --> 00:11:58.399 tool of this... Well, could we just explain why 00:11:58.399 --> 00:12:01.659 frames per second matter? Why they are the limitation? 00:12:02.059 --> 00:12:05.179 Sure, absolutely. Essentially, the software that 00:12:05.179 --> 00:12:09.019 is used in this in these applications essentially 00:12:09.019 --> 00:12:12.659 tracks the position of the plate frame by frame. 00:12:12.799 --> 00:12:16.100 So by tracking the position, it tracks the position 00:12:16.100 --> 00:12:18.059 of the plate and refers this position of the 00:12:18.059 --> 00:12:21.419 plate to an origin of a system of reference. 00:12:21.519 --> 00:12:24.480 So when the lifter is about to start, the plate 00:12:24.480 --> 00:12:26.899 is in the lowest position. So the app generally 00:12:26.899 --> 00:12:29.759 boxes the plate into a square and determines 00:12:29.759 --> 00:12:32.039 the center of the square. The center of the square 00:12:32.039 --> 00:12:34.200 would be the origin of the motion of the parallel. 00:12:34.379 --> 00:12:37.639 As the frames go by, the position of that square 00:12:37.639 --> 00:12:40.179 that identifies the position of the plate changes 00:12:40.179 --> 00:12:43.240 in space. So a new center of the plate is identified 00:12:43.240 --> 00:12:46.120 and obviously a new center has a different coordinate, 00:12:46.259 --> 00:12:48.799 different distance from the ground, right? That's 00:12:48.799 --> 00:12:50.879 what we want to measure. Now, the problem is 00:12:50.879 --> 00:12:54.419 if we don't have enough temporal resolution of 00:12:54.419 --> 00:12:57.639 the images, in other words, if we have only few 00:12:57.639 --> 00:13:02.919 images of the lift, we may miss the actual peak 00:13:02.919 --> 00:13:05.740 moment in which there is the maximum... extension 00:13:05.740 --> 00:13:08.419 of the barbell before the velocity starts to 00:13:08.419 --> 00:13:11.740 reduce. So essentially by increasing the number 00:13:11.740 --> 00:13:13.960 of frames per second that you can get, you may 00:13:13.960 --> 00:13:16.899 not, you would eventually reduce the possibility 00:13:16.899 --> 00:13:20.720 of making an underestimate or more overestimate 00:13:20.720 --> 00:13:23.820 of the actual velocity. So the camera is measuring 00:13:23.820 --> 00:13:28.500 here. The next photo is taken here. If the peak 00:13:28.500 --> 00:13:30.899 velocity happens here, that's missed. So when 00:13:30.899 --> 00:13:33.159 you have this and this. Exactly. You're essentially 00:13:33.159 --> 00:13:37.570 missing. Or the larger is the number of pictures 00:13:37.570 --> 00:13:40.129 that you can take, allows you to take a better, 00:13:40.330 --> 00:13:43.230 to a pricing with a higher precision, the moment 00:13:43.230 --> 00:13:46.070 in which you attain the peak velocity, essentially. 00:13:46.250 --> 00:13:48.750 So for the mean velocity, maybe it's not such 00:13:48.750 --> 00:13:51.210 a big problem because you kind of average all 00:13:51.210 --> 00:13:54.250 the measurements across the whole movement, like, 00:13:54.330 --> 00:13:57.110 for example, in squat. But for snatch and clean, 00:13:57.230 --> 00:13:59.830 where the movements are very fast, this loss 00:13:59.830 --> 00:14:03.690 of data is quite significant and will impact. 00:14:04.169 --> 00:14:06.990 accuracy. Yes, absolutely. I agree. And it's 00:14:06.990 --> 00:14:09.929 also a matter of, yes, absolutely. Okay. So Gmware, 00:14:10.090 --> 00:14:12.970 the most accurate, unfortunately the most expensive, 00:14:13.049 --> 00:14:15.629 then phone app probably would be more accurate 00:14:15.629 --> 00:14:18.889 if the camera would be better. So maybe in a 00:14:18.889 --> 00:14:21.289 few years we will get there where the general 00:14:21.289 --> 00:14:26.509 phone camera will be sampling in 120 or maybe 00:14:26.509 --> 00:14:28.789 more, I don't know. I believe that you should 00:14:28.789 --> 00:14:34.049 go for a higher than 120 probably. Yes, probably 00:14:34.049 --> 00:14:36.409 you should have. The good thing of the phone 00:14:36.409 --> 00:14:38.990 app is that you can also upload videos and analyze 00:14:38.990 --> 00:14:42.669 the videos. So, for instance, if you could acquire 00:14:42.669 --> 00:14:45.610 data with a GoPro camera that is pretty good 00:14:45.610 --> 00:14:48.409 at taking images with a very large number of 00:14:48.409 --> 00:14:52.269 frames per second, you could upload this video 00:14:52.269 --> 00:14:54.629 into your phone or tablet or whatever device 00:14:54.629 --> 00:14:57.190 you want to use and analyze the data with this 00:14:57.190 --> 00:14:59.950 phone application. Obviously, this defeats the 00:14:59.950 --> 00:15:02.549 purpose of the practical aspect of having a phone. 00:15:02.669 --> 00:15:04.830 the gym and measuring on the spot. So it makes 00:15:04.830 --> 00:15:08.409 further, it increases the difficulty of the logistic, 00:15:08.409 --> 00:15:11.649 if you want, the logistic complexity of implementing 00:15:11.649 --> 00:15:14.789 a parallel velocity training system. But you 00:15:14.789 --> 00:15:18.269 could set it up at the competition. So it's, 00:15:18.269 --> 00:15:22.149 you know, measuring maximal effort at the right 00:15:22.149 --> 00:15:24.389 environment, the optimal environment, which is 00:15:24.389 --> 00:15:26.710 a competition for a weightlifter without even 00:15:26.710 --> 00:15:28.909 touching the lifter. Absolutely. Yes. That's 00:15:28.909 --> 00:15:30.970 minimally invasive. I mean, it's non -invasive 00:15:30.970 --> 00:15:34.080 essentially. Yes. So I just wanted to move to 00:15:34.080 --> 00:15:38.480 Flex device and, oh yeah, these two. So could 00:15:38.480 --> 00:15:40.960 you comment on these? How did they perform in 00:15:40.960 --> 00:15:43.460 terms of accuracy and what were the problems? 00:15:43.820 --> 00:15:46.139 So the V -Trub also had a performance that was 00:15:46.139 --> 00:15:49.139 comparable to the phone app. And it's also true 00:15:49.139 --> 00:15:52.379 that V -Trub was the lowest cost position device 00:15:52.379 --> 00:15:55.659 of the three that we tested. Flex also had an 00:15:55.659 --> 00:15:59.059 error that was in the neighborhood of 0 .1 meter 00:15:59.059 --> 00:16:02.659 per second, a little bit more than. 0 .1 0 .12 00:16:02.659 --> 00:16:05.419 meter per second according to the type of lift 00:16:05.419 --> 00:16:08.039 probably gravity box was the next best device 00:16:08.039 --> 00:16:11.019 after the gym aware still included an error of 00:16:11.019 --> 00:16:13.820 0 .1 meter per second in the clean at the box 00:16:13.820 --> 00:16:20.000 and 0 .05 meters per second for a snatch so in 00:16:20.000 --> 00:16:22.740 my opinion still difficult to would be still 00:16:22.740 --> 00:16:26.440 problematic when you are interested in to determine 00:16:26.440 --> 00:16:29.639 the actual peak velocity at very high percentages 00:16:30.779 --> 00:16:33.279 Do you think if obviously many coaches, most 00:16:33.279 --> 00:16:35.399 of the coaches will not have access to motion 00:16:35.399 --> 00:16:39.860 capture. So this level of data accuracy is unattainable 00:16:39.860 --> 00:16:43.960 for a casual coach or the coach who is not working 00:16:43.960 --> 00:16:47.539 in a sports center or Olympics preparation center. 00:16:47.840 --> 00:16:51.919 So would it be still worth buying a linear transducers 00:16:51.919 --> 00:16:55.940 or they are basically at the level, apart from 00:16:55.940 --> 00:16:58.320 DreamAware, that is pretty accurate. Are they 00:16:58.320 --> 00:17:02.289 at the level of the mobile? How do they, from 00:17:02.289 --> 00:17:05.190 the practical sense, how do they line up? Well, 00:17:05.269 --> 00:17:07.650 I mean, from the practical sense, I think that 00:17:07.650 --> 00:17:11.029 besides Gemaware, that it showed to be extremely 00:17:11.029 --> 00:17:13.430 accurate compared to the others. These other 00:17:13.430 --> 00:17:16.289 linear transducers, in my opinion, are not, in 00:17:16.289 --> 00:17:19.240 my opinion, they are not designed in a way. Well, 00:17:19.319 --> 00:17:21.460 they could be certainly used for this purpose 00:17:21.460 --> 00:17:24.099 with all their limitations in accuracy. But I 00:17:24.099 --> 00:17:26.440 think that the main limitation is the risk that 00:17:26.440 --> 00:17:29.240 you can actually break them because of the nature 00:17:29.240 --> 00:17:31.859 of the type of lift that you do. So I wouldn't 00:17:31.859 --> 00:17:35.039 feel comfortable in a gym. If I were a gym owner, 00:17:35.220 --> 00:17:37.299 I wouldn't be comfortable in purchasing such 00:17:37.299 --> 00:17:39.880 a device with the high risk that this can be. 00:17:39.960 --> 00:17:42.319 It is very possible that you can break it. Me 00:17:42.319 --> 00:17:45.539 personally, as a user, as an athlete, I wouldn't 00:17:45.539 --> 00:17:48.500 make this type of investment personally. Okay, 00:17:48.500 --> 00:17:52.119 how about the Flex device, the laser optical 00:17:52.119 --> 00:17:56.140 one? Well, the Flex device actually still presented, 00:17:56.200 --> 00:17:58.680 in my opinion, still presented an error that 00:17:58.680 --> 00:18:03.089 is... almost comparable to the phone app. We 00:18:03.089 --> 00:18:06.529 are comparing, in terms of snatch, we are comparing 00:18:06.529 --> 00:18:12.069 an error that is 0 .1 versus 0 .1. We are pretty 00:18:12.069 --> 00:18:16.230 much there in terms of errors. So I wouldn't 00:18:16.230 --> 00:18:18.730 make that investment. It's a very cool tool, 00:18:18.829 --> 00:18:23.609 I think. The principle is very interesting. It's 00:18:23.609 --> 00:18:26.369 a very nice tool, but according to the data that 00:18:26.369 --> 00:18:29.750 we report, it's probably not the best. Do you 00:18:29.750 --> 00:18:32.250 think it is related, or maybe I will ask it differently, 00:18:32.470 --> 00:18:34.990 why do you think it didn't perform that well? 00:18:35.190 --> 00:18:39.250 I am not sure, honestly. I don't know if the 00:18:39.250 --> 00:18:43.250 problem is making sure that there is the right 00:18:43.250 --> 00:18:45.829 illumination, the surface presents wrinkles, 00:18:46.309 --> 00:18:48.549 the reflective surface, the matte that you place 00:18:48.549 --> 00:18:52.250 underneath presents wrinkles, or there are differences 00:18:52.250 --> 00:18:56.190 in the index of reflection within the matte, 00:18:56.309 --> 00:18:58.940 or there must be some dust. that changes this 00:18:58.940 --> 00:19:01.119 index of reflection and probably can change something. 00:19:01.160 --> 00:19:04.849 I don't know. honestly why he didn't perform 00:19:04.849 --> 00:19:09.410 in he didn't perform as expected at the end i 00:19:09.410 --> 00:19:12.430 always ask a question about the favorite lift 00:19:12.430 --> 00:19:15.289 i asked you before so if people are interested 00:19:15.289 --> 00:19:17.569 they have to go to the previous interview and 00:19:17.569 --> 00:19:20.829 i have a different question for you you are very 00:19:20.829 --> 00:19:23.650 well versed on both sides the research in weightlifting 00:19:23.650 --> 00:19:26.369 and the practical side you are an athlete and 00:19:26.369 --> 00:19:28.769 you try the velocity for programming and so on 00:19:28.769 --> 00:19:31.650 so you are a coach too now if you could change 00:19:31.650 --> 00:19:34.339 one think in Olympic weightlifting as a sport, 00:19:34.559 --> 00:19:36.940 what would it be? Wow. I think that the sport 00:19:36.940 --> 00:19:41.960 as it is conceived is very simple, right? Because 00:19:41.960 --> 00:19:45.799 it's probably, I would say, a better way of determining 00:19:45.799 --> 00:19:48.200 and an objective way to determine whether the 00:19:48.200 --> 00:19:51.519 lift is valid or not. I mean, we talk about pressing 00:19:51.519 --> 00:19:55.319 and position angle of the elbows in pressing, 00:19:55.500 --> 00:19:59.299 in snatch and in the jerk pace. If we could have 00:19:59.299 --> 00:20:02.480 a system implemented that that can objectively 00:20:02.480 --> 00:20:04.779 determine whether or not the link is valid. Well, 00:20:04.839 --> 00:20:07.400 there is actually a system. They could actually 00:20:07.400 --> 00:20:10.180 replay, but maybe in real time, I don't know. 00:20:10.299 --> 00:20:12.400 Yeah, just to make this experience a little bit 00:20:12.400 --> 00:20:14.799 transparent, more transparent for the audience, 00:20:14.960 --> 00:20:17.599 I guess. Yes. All right, fair point. Now, where 00:20:17.599 --> 00:20:19.920 people can find you if they want to follow your 00:20:19.920 --> 00:20:22.599 work or contact you about your research? Oh, 00:20:22.619 --> 00:20:26.680 well, they can look up our website, and we also 00:20:26.680 --> 00:20:29.500 have a Twitter account, and every time that we 00:20:29.500 --> 00:20:31.940 publish a new research, or there is a new thing 00:20:31.940 --> 00:20:34.859 going on in our lab, we always tweet and they 00:20:34.859 --> 00:20:37.660 can always consult for any type of information 00:20:37.660 --> 00:20:40.180 regarding us on the website of the University 00:20:40.180 --> 00:20:42.660 of Miami, Florida. And could you mention the 00:20:42.660 --> 00:20:45.099 handle for the Twitter account? Yes, that would 00:20:45.099 --> 00:20:48.359 be um underscore biomech. Awesome. Thank you, 00:20:48.380 --> 00:20:50.259 Francesco. It was a pleasure. It was my pleasure.