Retiring robots

Community General Discussions
1

I’m slowly getting back into robot building mode and have realized that I had started working on version 2 without leaving version one in a very good state.

So the other day I set about trying to breathe some life back into version 1. I of course had lots of unfinished plans for it that didn’t quite pan out (hence version 2), no working batteries, and the network had changed around enough that it was a hell of a chore just to get comms to it. Then I realized the current batch of software was geared towards a whole suite of things that it didn’t do very well simply because that’s the course of the development and the impetus for version 2.

There’s enough work in the thing that I want to leave it in a semi-functional/demoable state. I think that probably means basic walking around and demonstrating some of the PCT stuff and it’s limitations that led to version 2. In reality, it hadn’t walked around since very early in development and so the code that drove that is largely incompatible with the much more evolved code for basic postures and what not.

The crux of all this is that I’m going to have to spend a fair amount of effort (and some cash for a new LiPo battery) to let this bot retire with some class.

Anyone else tried to truncate a capability list and actually “finish” a robot before moving on?

2

um, my WiBRAT gave up his parts for my tracked rov and the ses scout/wi I have been slowly picking at after another recent servo and parts infusion. So, other than being a learning experience and stepping stool into other projects I’d say no… not so much left in a functional condition. :wink:

3

Yep, I always truncate the capability list before I finish the robot. :smiley:

4

What is this “finish” of which you speak? :confused:

Seriously, I tend to see most of my projects as open-ended testbeds, and I rarely have a set “finished state” in mind for them. When one becomes insufficient to serve as a basis for something, I’ll move it to a new/different chassis that can do it, or build/modify one for the purpose. Yes, most of my chassis have multiple sets of holes that serve no purpose in their current configuration, and are a hodgepodge of multiple generations of experiments layered on top of each other.

This is certainly not the most streamlined approach to robot design, but then, I’m not usually designing to a specific purpose, either. If I were to build a machine to compete in a contest with a set task, I’d most likely either design it from the ground up for that purpose, or pick a chassis that is well-suited to the task, strip it down to the basic structure, and then rebuild it specifically for the purpose.

Of course, anything that isn’t currently involved in a project becomes potential spare parts as well, but with an eye as to how difficult it would be to replace the bits being cannibalized should I decide to go back to that chassis at a later date. This approach has worked out well in the past, as one of my little desktop testbed bots that’s been used for sensor testing is now on its third generation brain board and its second display/interface module.

5

Seamus,

Very much the same thing happened with “version 1” which might just as aptly be named “version 26” considering all the bits that were rebuilt over and over as I started to figure out what the hell I was doing. What were original design goals had been met and surpassed long ago.

Yet this next change involves a essentially zero parts in common and could scarcely be called a revision. Switching from all PICs onboard to Gumstix and Atmel. Lexan to SES. Regular HiTecs to OpenServos (I have enough spare that I don’t need to cannibalize the servos for conversion). LiPo to (insert right answer here when decided). Regulated power to Unregulated.

So it truly is a scratch build leaving the previous version wholly obsolete.

6

Ussually projects I start are mostly apart of my science fair projects, therefore they always get complete to a set date. I think this is the reason why I always finish my projects and also becuase I hate to leave something unfinished, unless its for testing or hobby. Ofcourse I’ve only made 3 bots (presently on my third) and haven’t any plans in the future other than maybe a BRAT. If this is to happen, im sure after I get bored with it (is that possible with a BRAT :smiley: ?) he will end up as part of an Arm or something… :imp: :smiling_imp:

7

Oh well for me it’s just different. Because the robot is finished when it can be sold as a product for others to build, hack, modify, or improve. But it has to work first. The irony is my products help many people build their ultimate robot, but the process of actually making the products available often make it impossible for me to build my ultimate robot. :frowning: It’s fun anyway… :smiley:

8

Ultimate robot eh… :imp: :smiling_imp:

9

I don’t think you will get bored of the BRAT,Trust me,Its pretty fun :wink:

10

When I decide to build a robot, I do have at least a general idea of what I want the robot to do or accomplish. With WALTER (my first robot), I wanted to learn about sensors, general locomotion, etc, and a two wheeled robot seemed the least expensive way to get that knowledge and experience. At this point, I have put nearly as much money into WALTER as I would need to buy an LM kit, but the journey and what I have accomplished with WALTER has been and will always be worthwhile for me. :smiley: From my point of view, I think WALTER v1.0 has been a wonderful success and learning experience.

Now, I think I have outgrown what the current version of WALTER is capable of doing, so the next revision will have four wheels instead of two. As you might expect with my designs, I have a couple new twists of my own to add… :smiley::smiley:

I had a plan right from the start when I started designing The BiPod, and that plan remains. I just have to fill the assembled brackets with servos, get SEQ, and make it all wireless using Bluetooth (for starters).

I even have a plan for what I want Walk 'N Roll to accomplish, even though the final form it takes may end up being different from my current design.

8-Dale

11

this is one of the reasons why I believe that local robotics clubs are important to the hobby. Many of them have contests or demonstrations, whereby the members are given a challenge - some specific task to accomplish - and they have to come up with something that achieves that goal within the set constraints. This gives the members something to work towards, while hopefully leaving them open to experiment and play around with their own ideas.

There’s nothing wrong with having a goal of “I want to build a robot”, but I have often found that this leads to a fairly disorganized approach, with a bunch of sensors or other features that are barely implemented, and a generalized machine that doesn’t do a whole lot of anything very well, and will probably never actually be “finished”.

By having a specific goal such as, for example, “Enter a defined area, scout for and collect the objects found there, and return them to a collection point, within a three-minute timeframe”, you have a well-defined task to accomplish, and can build a machine that does it exceptionally well, without a lot of extra bells and whistles that just load down its power, cargo, and computing capacity. Once the club has its “big meeting” where the contest is run, and the next contest is announced, then you have the option of rebuilding your machine to achieve the new task, building a new one to suit, or modifying your existing one to incorporate both tasks in its “bag of tricks”.

Many clubs have a number of different contests, with increasing levels of difficulty, that they run repeatedly. This allows new members to progress up the scale at their own pace, and with ever-progressing technology, it’s a fairly certain bet that someone’s new bot that does well on the line-following obstacle course will look very little like the champion from just a few contests ago.

Developing one’s own generalist robot is great - it lets you pursue your own interests at your own pace, and come up with some really fantastic stuff. Club events and contests are also important though, as they can provide focus and a set goal, which can help to offset the robotic equivalent of ‘urban sprawl’, and result in some very innovative solutions to problems, which can then be applied to other areas as well.

I for one feel that I’d be much farther along in my own robotics progress if I had had robotics clubs available, with meetings that my schedule would allow me to attend. I read about all these great contests and productive exchanges of ideas, and my mind boggles over how much people are learning from each other, even just by watching someone else’s robot make its way through an obstacle-avoidance drill.

12

One ultimate purpose I have for a robot I will eventually build is to compete and at least finish the course in Robo Magellan. :slight_smile: I’m still learning what I will need to design such a robot so am still far from even designing such a robot. I won’t count such a robot as a success unless it can actually finish the course, and it doesn’t have to win although that would definitely be a nice plus. :wink:

8-Dale