Hello, I am new to the site and am very happy to have found it. I am an electrical engineering student at NYU Polytechnic and am looking for some advice on a project I am participating in.

The project is a lunar excavator for a NASA competition that must be controlled via a router to navigate an obstacle course, mine lunar regolith simulant and return it to a specified area. I am on the controls subcommittee for the project and we are currently in the research phase. The first step is choosing between using a nettop or microcontroller and the choice must fit with the objective of the competition. Said objectives include:

1. Minimal weight - the robot must be under 176 lbs and every little bit counts so much so that we will probably end up having to scrap parts to make weight).
2. Minimal bandwidth - points are allotted for different categories one of which is bandwidth used. The maximum bandwidth allowed is 5 mb/s and some teams have come in under 1 mb/s. This is not total bandwidth but average over the time of the collection so we can have peaks and troughs.
3. Minimal cost - we are, after all, a university team trying to get our own funding.
4. Ease of use - not everyone on the team is proficient in microcontrollers even more so nettops and those that are will not necessarily have experience in the specific one that fits the job. It would be preferential that the selection be easy to learn although I should point out we are quick learners with plenty of people to turn to for assistance.

As it stands my research has pointed to benefits in both. The nettop has more features; onboard camera, wifi, larger memory which are needed for navigating the obstacle course. A particular plus is the memory to program macros that will minimize the information we have to send if that message has a much larger command preprogrammed on board. The microcontroller has the advantage of smaller weight and power requirements as well as less unnecessary features which the nettop will invariably come with.

I could really use some advice as well as specific suggestions. Currently I am leaning on the side of the nettop but will be considering all advice and we will ultimately put the matter to a vote.

Thank you so much in advance,
Yusif

 

As a Poly alumni I of course have to weigh in with whatever I can here.

You've laid out two good choices, a minimal "PC" platform vs a microcontroller. I would only add there is a "PC-104" family of devices which the PC form factor is held down to a series of stack-able boards under 4" per side. They are widely supported and have numerous peripheral boards available as you will see in that article. Even if your budget doesn't cover them perhaps you can beg for donations?

I prefer to let my constraints dictate my hardware choices as must as I can so I have to ask how hard are your processor requirements nailed down? Do you need "onboard camera, wifi, larger memory" or are you still sketching out the system?

Micros can be incredibly powerful, I am always amazed at all the information the tiny GPS unit on my dash board has, every address on every road along with individual location tags including store name, address, phone number, even an icon! Honestly I do not understand how all that information is packed in there, but even very simple micros can access SD cards and get gigabytes of local information and storage.

As both a PC and a micro can run C code you may delay the choice of final platforms as you work out what code is required to run the bot then customize it in the final stages. Make sure it is written in layers so the "application" layer has no knowledge of the "driver" layer so you can swap out PC for micro routines.

Just some thoughts, I do not have a fraction of the information needed to make but general comments.

 

I would have to go along with ErnieM. You don't let the hardware dictate the system. Decide first what functions you need followed by what software will be required. Let the software determine the hardware.

 

How would the nettop interface with the various sensors & actuators on the robot? I have never built a robot, but if I were going to (and I wanted it to be controlled by a PC) I would probably end up using a microcontroller as an interface beween the field devices and the PC. If the microcontroller is going to be there as in interface, then it might as well be the big kahuna - get rid of the laptop. Just my uneducated opinion; there are likely better ways to interface a laptop to sensors & actuators than a microcontroller.

 

You may want to take a look at this:

http://blogs.freescale.com/2011/06/...rd-i-mx53-quick-start-board-and-vex-robotics/

 

Guys first I just want to say thank you so much for your responses! I am truly grateful that you would take the time to help me out with this as I am only beginning to get my feet wet controls-wise. Let me first address the follow up questions you posted:

ErnieM: Yes, we will be soliciting donations the whole way! As a matter of fact the entire endeavor is an exercise in solicitation from asking departments for affiliation and a work station to asking big companies for funds to get parts and tools. Will definitely look into the PC-104 you suggested though I am as of now not familiar with it.

The things I mentioned "camera, wifi, memory" are indeed things we need for the robot; camera to see where it is going, wifi to control via the router we are provided and larger memory to program macros i.e: left two motors go forward until stop, right motors go back until stop all in one command memorized onboard. This way the bandwidth would be lower when sending a command. These were the pluses on an nettop while there are minuses like extra programs running that we do not need, weight, power, etc.

ErnieM & MrChips: I see what you guys are saying however the controls on this thing will be pretty minimal; two links pairs of motors will essentially have a forward and back options (I/O) as well as speed control. There will also be on/off settings for the ogres and several more interaction commands as well as those I am not thinking of right now but ultimately there will mostly be on/off commands that need to be communicated wirelessly using the least bandwidth. The design of the robot is really the difficult thing because the frame is essentially the innovative part. As I mentioned this is a multidisciplinary project so the electrical part, while very important, is not the "meat" as it were. I am still pretty green in my studies not to mention such large scale projects so I need all the help I can get.

Strantor: As I was advised by a professor and our resident robotics genius, the easiest way would be to use a microcontroller to interface the nettop with the mechanism. I guess the question is morphing to whether or not the nettop should be used at all or if it should be just the microcontroller. Another professor thought it would be enough though I am considering the advise of starting with microcontrollers and should I reach the limit of their ability, say with regard to memory, add the nettop if necessary.

Thank you again for your help fellas. I realize, being new to this site, I may not be asking the right question but I am essentially trying to figure out which would be the best place to start. Whatever we go with, we will buy the piece and begin getting aquainted until the end of summer at which point we will start simulating the controls with LEDs or something to that effect. In my mind this thing should be a tank drive with speed controls which seems to not require an extravagant amount of commands but then again I could be wrong. My ideal scenario would be to hook up two joysticks to the controlling laptop and drive the thing like a tank although the project lead is skeptical.

Once again any advice is appreciated and I will try my best to answer any follow up questions to make it easier for you. I hope to keep this thread going throughout the construction of the project and ultimately share pictures of the finished product.

 

Strantor: As I was advised by a professor and our resident robotics genius, the easiest way would be to use a microcontroller to interface the nettop with the mechanism.

Cool; feels good to be validated by professor and robotics genious. Here I thought I might be talking out of my rear end.

 

Thank you for all the advice fellas! I just wanted to give you an update on the decision that was reached; we will be using Arduino Microcontrollers with the option of adding a netbook should we run into the mictrocontrollers' limitations. The majority of the opinion, here, speaking with robotics enthusiasts at school and embedded systems professors has been towards just using the microcontrollers and forgetting about the netbook. A digital logic professor I had suggested a board with FPGA (for networking) and Microcontroller chips onboard and we threw around that idea for a while but when we spoke to the embedded systems professor he was pretty confident we should just use an Arduino with networking capability.

With that decision out of the way I am now getting acquainted with the Arduino and will be looking into network add ons shortly.