>>939392 Yank out all the ultrasonic sensors, they're shit. Put in a kinect, depth camera, or neato laser scanner, then put in something that can run Robot Operating System(ROS) and process that shit. Then do SLAM. If it's not doing SLAM then it's barely a robot
>>939485 >that shovel is shit >remove the steel shovel head and replace it with this titanium model >while your at it remove wooden the handle from the shovel and replace it with this fiberglass handle
>>939420 >>939485 Hmm... what if instead of that OP hacked his HC-SR04s to circumvent the onboard processor, then arranged the three of them into a phased array for generating a 3D point cloud of his surroundings? http://hackaday.com/2015/12/04/el-cheapo-phased-array-sonar/ You could maybe even use synthetic-aperture shit referenced with the odometry of the rover to improve resolution.
>>939392 I used the exact tamiya chassis for my undergraduate design project. It had one of these with a stepper motor under it for rotating it http://www.ebay.com/itm/Mini-Hidden-Pinhole-Micro-Spy-Cam-Complete-System-Wireless-Security-Nanny-Camera-/290940950784 I did the vision and navigation part on a computer, then send the movement commands with RF.
One annoying thing was that whenever motors ran (L293D driven with PWM), I couldn't get anything from the RF. I tried shielding the motors, all kinds of caps but couldn't get the thing to work, so I had to shut the motors off once in a while to listen to the RF. It would be awesome if someone can tell me what I did wrong.
>>940534 well there's this short range one: http://www.aliexpress.com/item/1PCS-VL6180X-Time-of-Flight-Distance-Ambient-Light-Sensor-VL6180-High-accuracy-Range-finder-optical-ranging/32579207577.html?ws_ab_test=searchweb201556_2,searchweb201644_3_505_506_503_504_502_10014_10001_10002_10016_10005_10006_10003_10004,searchweb201560_1,searchweb1451318400_-1,searchweb1451318411_6452&btsid=03f4f774-5d88-4b16-870a-7ceba110bf52
I don't know if I'd trust it much beyond 25 cm.
This isn't actually a laser, it's an infrared proximity sensor, range is 80 cm, it's analog and nonlinear though: http://www.aliexpress.com/item/2015-New-GP2Y0A21YK0F-GP2Y0A21-Infrared-Proximity-Sensor-IR-Analog-Distance-Sensor-VE713-P/32431180659.html?ws_ab_test=searchweb201556_2,searchweb201644_3_505_506_503_504_502_10014_10001_10002_10016_10005_10006_10003_10004,searchweb201560_1,searchweb1451318400_-1,searchweb1451318411_6452&btsid=bd51d5b4-fcff-453e-8bca-d4c005f3b214
expensive but really fucking good(requires you scan it around though): http://www.teraranger.com/product/teraranger-one-distance-sensor-for-drones-and-robotics/
Neato xv-11 lidars can be acquired pretty cheap on ebay and are pretty fucking good.
>>940488 >Well you can certainly use a sonar to get a 3d point cloud, it's just that you need an acoustic lense and a custom made transducer array, way out of OP's budget. Firstly, it's not "and," it's either-or.
Secondly, bats seem to navigate and hunt just fine with only two "transducers." I think an array of three could be plenty. Plus, with some DSP referenced to the motion of the bot and sonar returns taken from different positions, you could effectively synthesize a much larger aperture/array and thereby produce a fairly high-resolution image of the surroundings.
>>940822 >Well if it is so simple, then why don't you make one? Oh, I intend to. I'm still researching the necessary DSP and haven't yet decided whether it will be ultrasonic or radar, but I intend to build a synthetic aperture scanner, and HOPEFULLY integrate it into my UAV. Should be a good learning experience. >If you make something with two transducers Two? Hell, why not one and a duplexer? >that can produce a high resolution image of the surrounding Relatively high-resolution, as far as ultrasonic/sonar goes. I don't mean to imply it's the most precise spatial scanning technology out there, sorry if I gave you that impression. >you could make a bunch of money. For doing something that's already been done before? Well, maybe. What kind of market would there be for such a product?
>>941152 ok I think I understand you want to do synthetic aperture stuff. This might work for a ground robot, if you had really good odometry. Unfortunately, you can't get really good odometry. Whenever pretty much any robot turns, there is a pretty large uncertainty in how much you actually turned. So you use a laser scanner or whatever to look at your environment to correct these sorts of problems.
>>For doing something that's already been done before? we don't know how bats' sonar works.
>>what kind of market well for one the navy. Second, an obstacle avoidance system for UAVs.
Now on bat sonar. Bats can't process signals as fast as DSP does, so we are pretty sure they do a good portion of the signal processing 'mechanically'. We know that they have a weird shape nose that sends different frequencies to different places. For example, we know one bat species, that flies in open plains, it's low frequency sounds downward and its high frequency components upward. Low frequency probably tells it where the ground is and high frequency probably tells it where food is. Same applies for their ears. Pic related is the crazy radiation pattern from a bat ear.
Not to mention that they scan their ears and nose around.
>>941176 >Whenever pretty much any robot turns, there is a pretty large uncertainty in how much you actually turned. A magnetometer might help. >we don't know how bats' sonar works. I wasn't exactly talking about bat echolocation there, but there is a bit known about it. >for one the navy. Pretty sure they have people far more qualified than I for this sort of stuff. >obstacle avoidance system for UAVs. SAR is kinda too slow for obstacle avoidance, but simple pulse-doppler or FMCW sonar could be useful. >Bats can't process signals as fast as DSP does The only "fast" thing about SAR DSP is the initial digitization. Most of the DSP is post-processing, after many signals and returns have been collected. >For example, we know one bat species, that flies in open plains, it's low frequency sounds downward and its high frequency components upward. Interesting... using frequency diversity on the transmit beams to distinguish between received returns. Reminds me of how monopulse radar exploits polarization of EM waves for similar purposes. >Not to mention that they scan their ears and nose around. Kinda hard to employ phased-array transmit steering with only one larynx. In fact, with such a small nose/mouth, I doubt the vocalizations are really THAT directional at all. They DO use phase/time difference between their two ears to perform sound localization though, as do humans. Head-turning isn't necessary for that.
>>939485 The kinnect sensor requires a LOT of RAM and computation on the host, as well a USB port. The ultrasonics require nearly no computational power, a handful of bytes of RAM, less than ten lines of code, and maybe two GPIO pins per.
If they're enough to handle the needs, it's a great way to handle them.
>>939687 I was using locally made RF modules, but they were well made and expensive (I upgraded assuming it was the cheap modules). So I don't think it's the modules.
Obviously there is a single battery on the device, so they are using the same power source. I used all kinds of electrolytic caps. In >>939659, the sideways cylinder behind the camera is the largest capacitor I could find. I didn't try any other types or more complex stuff. I believe the crystal filter you mentioned is probably already used in the RF module.
I figure there is an obvious design rule that every person that uses a motor and an RF device on the same circuit knows about.
I am still looking for answers so let me repeat what the problem is: >microcontroller (pic18f4550) is driving the motor with L293D >to control speed, enable pin is toggled by PWM >microcontroller works fine while the motor is running >microcontroller using an RF module to get one byte instructions from the PC (w/ UART) >RF works fine when the motor is running >as long as motor is running, microcontroller receives nothing from the RF module. when the motor is shut off, communication resumes. >I tried putting a huge electrolytic cap near the RF module. >I shielded the gearbox with copper plates and grounded it. >using separate power sources is not an option. furthermore, I remember using a separate 9V battery and having the same problem, but I'm not 100% sure.
I'm considering some possibilities feel free to add please: >using PWM and UART simultaneously is not possible - the manual doesn't mention it. can't see people complaining about it on the forums so unlikely. >I should be filtering the connection between the microcontroller and the RF module - don't know if this is even a thing or how to do it
>>941237 Well guess what? Because of cellphones computation in a small form factor is cheap! A raspberry pi 2 has enough processing power to do SLAM using a kinect: https://hackaday.io/project/8629-poor-mans-raspberry-pi-turtlebot
All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the shown content originated from that site. This means that 4Archive shows their content, archived. If you need information for a Poster - contact them.
If a post contains personal/copyrighted/illegal content, then use the post's [Report] link! If a post is not removed within 24h contact me at email@example.com with the post's information.