Sunday, November 18, 2012

Running Dog Walk Training: Modification One


I still have other material on other subjects related our foundation training to write about, but, in the meantime, I added a modification to my running dog walk training set-up in order to help with one of the problems I've been having that wrote about in my last post. The issue is that Hokey often comes off the side of the plank as she runs toward a remote controlled treat dispenser placed several feet from the end. A friend of mine (thanks Ivette!) suggested using a hoop at the contact end that could act as a visual aid to keep her running straight on the plank all the way to the end. I do have a hoop, but thought that maybe just a couple of upright posts connected by a short stabilization bar (in this case, roughly 16") might be enough to do the trick.

First, I needed to introduce her to the concept of running through the uprights. I shaped this simply by strolling by the obstacle and rewarding her for taking the initiative of going through the uprights. We've done this when introducing other obstacles in the past, so it's a familiar drill to her. The only "new" concept was the narrowness of the uprights. Here I am shaping the behavior. As you can see, as Hokey gradually becomes more confident about performing the behavior, her speed and drive increase. By the way, that soft click you might hear is a flashlight that I'm using as a substitute clicker. More about that later.

video


Once Hokey clearly understood the concept of running through the uprights, I placed the remote controlled treat dispenser on the ground for her to drive to. She would get a treat released for a "hit" - running through the uprights, but, in theory, wouldn't get one for a "miss" - running by the uprights without going through them. She never did get a miss during this phase though.

video


Then it was time to put it all together by putting my foam tiles down with the uprights placed at the contact end and the treat dispenser placed several feet from the end.















I have decided it is easiest to see a contact hit if I have 3 foam tiles of all the same color linked together at each end of the "plank" (recall from the last post that 3 tiles are equal to the USDAA contact zone measurement). Here I have red on one end, yellow on the other.




The modification with the uprights worked really well. It seemed to do its job of acting as a visual guide to help her stay on the plank all the way to the end the majority of the time. She had a couple of instances where she came off and then swung back on to run through the uprights, but those misses were not nearly as numerous as those during the training session a few days ago when I wasn't using the uprights. Here is Hokey getting some nice "hits":

video

As for the other main problem mentioned in my last post - that of the absence of a good clicker substitute to mark the desired behavior - that is something I'm still working on.

 As you can see from these stills from videos taken during today's session, she still has times where she is leaping over the contact area rather than hitting it.
 
There is a slight possibility that the uprights will encourage this behavior since she might interpret them as a cue to jump, but because she had several instances of just running through the uprights rather than jumping, I do not think this is the case.

Here is the video showing her leaping over the contact zone:

video


Without a proper aid to give her instant feedback, there is really no good way for me to communicate what I'm looking for her to do. I'm sure it's frustrating to her. Not having the ability to communicate my criteria to her properly is frustrating to me.

However, I do think I may have a workable solution. I bought a little push button LED flashlight a couple of days ago to try to use as a substitute clicker (as an alternative to my hand flash). It's relatively bright, so as long as Hokey is looking at it, she should be able to see the light flash on even when outside during the day. It also has a relatively quick reaction without much of a delay, which is a huge bonus as this is something I have struggled with when trying to find devices that will work well in our training.


The downside to using the flashlight is that Hokey needs to be facing it in order to get the information. In training the running dog walk, if I'm restraining her or holding a stationary position at any location other than right at the end of the plank facing toward her, I can't provide her with the information contained in that flash of light. Even with me standing at the end like that, things don't work out so well. Having to release her, watch the contact, decide to click the flashlight followed by clicking the remote for the treat dispenser all within quick succession did not prove to be very workable in practice. Therefore, I think it would be best at this time to have a second person stationed at the contact end of the plank watching the contact area for "hits" and clicking the flashlight, while I concentrate on releasing her and operating the treat dispenser.

In the short time I've introduced her to the concept of the flashlight as a substitute clicker (I picked it up on Friday and today is Sunday), Hokey has grasped the concept very well. Here we are today demonstrating how it can be used to shape a sit/stay. Notice that she is definitely glancing at the flashlight in anticipation of getting information/feedback from it.

video
 

So that's the update! Running dog walk contacts are still a work in progress, but with improvements!






Wednesday, November 14, 2012

Foundations: Running Contacts


I finally have time to report on the progress I've been making with Hokey's agility foundation training. In fact, I have so much material, I'm going to have to split it into more than one post. This post will be dedicated to our attempts to develop solid running contact behaviors for both the A-frame and the dog walk. 

But first, let me report that we all survived "Superstorm Sandy" (I still prefer the pre-storm moniker "Frankenstorm") without any hardship. However, Hokey begs to differ with that assessment. She was convinced she'd melt by getting rain in her face. Is this not the most pathetic sight ever?

Hokey "suffering" the effects of Superstorm Sandy


So back to running contact training starting with the A-frame. I have introduced Hokey to the PVC box. The introductory step involved me sitting in a chair while shaping her moment into the box (i.e. first moving toward it, then a foot in, two feet in, etc) by building value for it with lots of tossed treats. She moved through those steps very quickly. Then I ran into some trouble at what would ordinarily be the next couple of steps - me still sitting in a chair off to the side of the box with her moving straight through without looking at me while getting a nice pounce into the box then bouncing straight out. Because of her deafness, I had trouble imparting this concept to her and actually ended up jumping ahead a couple of steps. The main problem is that I have yet to come up with a good alternative marker to use in lieu of a clicker.  Such a marker would need to be visual, yet not require her to look at me for information (i.e. some remote device that she would look at to get her information regardless of my position relative to her). It would also have to have very little delay so that the feedback would be virtually instantaneous. I've experimented with a couple of things, but none of them have worked out quite to my specifications. SO, I decided to jump ahead and see how she would respond to me working the 3 positions around either side of the box (front corner, side, back corner) with me tossing a treat or toy to keep her focused on driving forward. Being a bouncy jack, she naturally picked up on the one bounce into the box with a bounce straight out. She really gets some nice pouncing action, as seen in this still from the video that follows:

Hokey "pouncing" into the PVC box

And here is the video of us working the 3 positions around the box at full speed followed by half speed:

video


We've also started to work on the foundation for a running dog walk contact. I want to try to train this and see how it goes. So far, it's proving to be challenging and sometimes frustrating. I may need to rethink how I go about certain things, but I figure if it doesn't work out, I can always switch to a 2o/2o instead. I think it would be an easier transition to retrain a running to a 2o/2o contact behavior than the other way around. Currently I am using interlocking foam tiles as a portable substitute for a plank. They are 12" x 12", so 12 tiles represents a DW ramp and the last 3 tiles are equal to the USDAA contact zone specifications.
One of the interlocking foam tiles

When training a running DW contact, you want to teach your dog to drive forward with speed while remaining completely independent of handler motion. Therefore, in these early stages of training, I remain stationary and position myself at either end of the "plank". I have a remote controlled treat dispenser placed several feet from the end of the plank that acts as the "goal" for Hokey to drive to. I've also used a tug toy stuffed with hot dogs placed or thrown at the contact end of the plank. However, if I use the toy, I give up the control of rewarding her for a correct vs. incorrect performance since she's often driving forward far ahead of me. In other words, the toy reward is indiscriminate; it rewards her for performing both correct and incorrect contact behaviors. Therefore, I prefer to use the treat dispenser so that I can reward her only when she runs all the way across the plank without leaping off the end or stepping off the side. The downside to using the remote dispenser is that it has a little bit of a delay, so that sometimes in my eagerness to deliver the reward to her at the correct moment, I anticipate her performing correct behavior and instead end up rewarding her for an undesirable behavior.

Remote controlled treat dispenser
Unfortunately, as mentioned before, I haven't come up with a good clicker substitute with which to provide her with instant feedback for performing the correct behavior without her having to look at me, so, right now, Hokey is stuck with the imprecision of trying to learn what gets her a reward and what doesn't and trying to tease out what she did right from the wide range of possible behaviors. This is a currently source of frustration for the both of us and something I need to give more thought to in order to come up with a better solution.

Below are video examples of our hits and misses.

In this clip, I stand stationary at the "contact" end of of the plank near the treat dispenser and release her toward me:

video


I have found that she tends to show a little more speed and drive if I am at the opposite end with her and do a restrained release, sending her across the plank toward the dispenser, as shown here:

video


As I mentioned before, Hokey is a bouncy jack russell and has a tendency to add an occasional leap amid her strides as she runs. Sometimes these leaps carry her right into the contact zone, such as here:

video

However, this is not the most desirable behavior as it can easily turn into a leap that has her sailing completely over the contact zone:

video


A frequent problem I'm having with her is that she has a tendency to run off the side of the plank as she's heading toward the treat dispenser.

video

This is happening more often lately and I need to fix it now before it drifts into habit. My thought is that I either need to take some steps back and build high value for staying on the plank before running the entire length at full speed OR get a helper so that one person can be available to give some sort of visual "click" for the desired behavior. Or maybe some kind of combination of the two.

Like I said, it's a work in progress with some kinks to be worked out. We'll see how it goes as we move forward.

Another post on our progress focusing on other skills should be up within the next few days. Stay tuned!

Tuesday, November 13, 2012

Extension & Collection: A Photo Study

As I've been editing videos for my next post (which I hope to get up tomorrow night), I thought it would be interesting to analyze the different responses Hokey has when reading my cues telling her to jump with either extension or collection.

First extension:

Exiting tunnel at full speed
 
Approaching jump at full extended run

Preparing for take off at full run. Hind legs extended in front of forelegs.

Pushing off fully extended

Lift off with hind legs fully extended

Extension over jump. Tail and rear legs extended.

More extension over jump

Landing. Looking forward.

Landing. Hind legs coming forward.

Landing Complete. Rear legs again coming in front of forelegs to push off into full run.

Then, jumping with collection:

Exiting tunnel at full run, as before

Responding to my collection cue by starting to check her stride. Notice how the back legs are not as far forward as when she is approaching in extension

Preparing for take off with collection. Shifting weight to hind end.

Take off with collection

Starting to turn. Notice the hind legs are much more tucked than when she jumps with extension. The tail comes down to help her steer into the turn.

Landing under collection while turning

Landing complete. Following through to complete turn.

Rear tucked in tight turn

Turn complete. Notice position of hind legs, collected under body with weight shifted onto them.
One more look at jumping with collection from another angle:
Approaching extended, but just preparing to check stride

Checking stride on approach

Checking stride again, preparing for take off

Starting to take off with collection. Notice the position of the rear legs and how her weight is shifted back.

A collected take off. Starting to turn.

Collected and turning in the air. Notice how the rear legs are tucked and the tail is down.

Following through with a tight, collected turn in the air.

Landed and completing turn on the ground.