HNHacker News
TopNewBestAskShowJobs

TFortunato

778 karma · joined January 23, 2011

I build robots.

Current: Staff Robotics Engineer at Chewy

Prior: Atlas @ Boston Dynamic; Robotics and Embedded Systems Engineering @ Amazon Robotics, Rethink Robotics, iRobot; Automation Engineering @ Ginkgo Bioworks.

Early Career: Electrical Engineer in Semiconductor Industry @ Applied Materials, Analog Devices, Allegro Microsystems

submissionscomments
TFortunato··on Ask HN: What mistakes in your experience does management keep making?
In this example, it states that because of this enployees were rewarded for reducing the unit cost of the product, which meant cutting in to quality. This led to sales declining because custoners were less interested in ab inferior product.

If management had looked less at the "how much profit can we squeeze out of 1 gallon" angle, and more "how much profit can we make overall", they might have made different decisions, such as preserving the quality of the product, they may have done better overall.

Unit profit is one factor in overall profit / revenue, but # of units solds is equally as important...it's a balancing act.

TFortunato··on Metallic Glass Gears for NASA Robots
Just wanted to add, that is a cool project on hack-a-day, but once you start looking at precision motion, or variable loads (eg. moving a robot arm), you start to become keenly aware of the physical imperfections of your motor and gear-box, and start having to think of things like cogging in the motor, or torque-ripple in the gearbox output and all that. The drive signals on those direct drive robots are NOT going to look like the perfect "3 sine-waves 120 deg out of phase" you see on wikipedia, and you need some damn good control software, and sensing of motor state to get good motion.

I'm not trying to say it's impossible, or to knock anyones project or sounds argumentative. It's been done, as you pointed out! Just wanted to point out that while in theory it's much nicer to be able to do direct torque control on the motors, the headaches to become apparent in practice. (I still love the idea of direct drive :-D )

TFortunato··on Metallic Glass Gears for NASA Robots
Haha -- yes indeed about the engineering headaches! (Although as I'm sure you know...the more I learn about any field of engineering, the more I learn about just how-many trade-offs there are in building something!)

As for gear ratio -- depends on which joint, but Harmonic Drive's standard gearboxes come in ranges of 50:1 to 320:1, and I'd say they were all within that range.

As far as direct drive motors go, all I can say is that I was very aware of them, and very interested in looking at it. (I was on the embedded systems team doing motor control), but by the time I joined, the architecture had already been decided and robot's were being built. That said, there are certain other difficulties there, both with the control electronics for larger motors, and with the relative cost of the motors, especially since Rethink is going for the goal of reducing robot costs. I can't say they won't go that way some day. I think at the very least, it's worth exploring, but I'm not longer there so I don't have much say in that!

Other fun stuff -- Even though we used a harmonic drive, it was coupled to what is basically a torsion spring between it and the output. This arrangement is called a series-elastic actuator, and it allows you to do some neat stuff with rapid torque sensing, both from the control and safety side of things. It's also a lot of fun for the control engineers trying to control something sitting at the end of a spring :-P

TFortunato··on Metallic Glass Gears for NASA Robots
If I recall correctly (Not an ME - I'm more of an electrical / software side of the robot guy) -- between the flexibility of the spline gear giving a tight engagement, and the high preload / friction at the gear interface, you end up just backdriving the input if you try to jerk around the output shaft.

They backlash may not be truly 0 (ask a knowledgable Mech E.), but they are definitely much better on that front than any other gearset with the same reduction I've ever got my hands on.

TFortunato··on Metallic Glass Gears for NASA Robots
Perhaps for some larger robots, but I can say that Universal Robots UR5s and Rethink Robotics Sawyers use them, so not completely out of favor :-)

Also the sets made by Harmonic Drive (the company) are definitely designed to be back-drivable. Your conment made me smile, because I remember that I heard the same thing about these gears. My first week at Rethink I asked my boss about it while looking on in slight horror at him moving around an unpowered robot while he was explaining the system to me -- I was very concerned we were about to break something!

TFortunato··on Metallic Glass Gears for NASA Robots
As another interesting aside on these gears -- besides the very cool materials in these particular gears, the technology behind "strain-wave" / "harmonic drive" gears used in a lot of precision robots is very very cool. Parts of the gear set are actually flexible yet you get very smooth* motion with negligible backlash out of them!

http://www.harmonicdrive.net/technology

(EDIT: We commented at same time, so see also olivierlacan's commented video! https://www.youtube.com/watch?v=bzRh672peNk )

* - (Compared to more typical gears, and the cogging you get out of them. There are other issues with these gears the controls guys have to deal with when you start going for really high precision motion, but I don't want to seem like I'm knocking them as they do solve a lot of problems in a really clever way!)

TFortunato··on CEO says iRobot will never sell your data
Totally understood - skepticism is always warranted when it comes to this sort of thing!

From my end I always add a disclosure about any current and former employers in when I post anything about them because I'd rather just be honest that I may be biased on the subject!

FWIW, I no longer work there, nor do I own any stock - My only interest in defending them is I have great memories of working there, and a lot of friends who are still there! (Again, feel free to take with a grain of salt)

TFortunato··on CEO says iRobot will never sell your data
Hey jal, just to let you know the iRobot privacy policy and device data collection and retention policy does state that data will not be shared, divulged or sold to any third party without your express informed consent on a case by case basis, and that data is not even collected by them unless you opt-in and register your robot with them. The bots are fully functional as a standalone device without internet connectivity. Additionally you can contact them at any time to have your private data removed from their servers which they are happy to do.

Full disclosure: I am a former iRobot engineer, and also a big privacy proponent. I also had some feelings about a camera enabled mapping robot when I first learned about the project, but I can tell you that the people there take privacy and data security VERY seriously, and really are trying to respect their customers as much as possible. Unfortunatley the message that iRobot wanted to allow you to share your mapping data if you wanted to for enabling smart home applications got spun pretty quickly into "iRobot wants to sell your data!")

TFortunato··on CEO says iRobot will never sell your data
To be honest I'm not sure, PR, marketing and such is not my strong suit. (Also, just wanted to add that any insider knowledge I share is solely on iRobot culture, and I cant divulge any insider info on technology, only whats publicly available. Sorry for the disclaimer, but this account is tied to my real name, haha!)

If anything I would say that there is some good information on the webpage they need to make sure gets out there. I'd also say that they should get out there and talk about their vision for the future more. Colin has talked a lot about his vision of a smart home, and how iRobot can bring a lot to the table by providing users a map of their home that is updated fairly often. As such I think they could get people more excited by not just saying "no we arent selling your data", but by showing the potential. Imagine being able to say "OK google, vacuum all the bedrooms and mop the bathroom", and having it dispatch your roomba and braava. Or saying "Alexa, have you seen the tv remote?" and having it respond "the roomba spotted it inder the couch when cleaning earlier".

(Not necessarily saying this is what is being worked on, again just my impressions I got that Colin / the leadership there is more interested in making cool bots and working to make a true smart home a reality, and willing to let users have access and share the kind of data a mapping robot can bring, and less about becoming a data/ad broker.)

TFortunato··on CEO says iRobot will never sell your data
Fair enough, though I do think calling it put as a lie is a bit strong, in the context of the FAQ linked since I read it as third-parties meaning other private companies, not government agencies in response to a legal request. (That may be my bias since I worked there, know the culture, and still have plebty of friends there).

I can absolutely see your point, and would say that they do have actual legal pages on their privacy policy and data security measure, which mentions that they will respond to a legal subpoena, and that you can at any time request that your personal data be deleted, which theu will do.

Again, look you can think I'm full of it amd take this with a huge grain of salt, since I did worm there. All I can tell you is that the people who worked there and started that company are there to make cool robots, not to be personal data brokers or an advertising company, and any conversations we had when I was there were about how we could make things work with other services and do neat smart home stuff, while keeping peoples information safe and privacy intact, not about how we can make money off of peoples data.

(apologies for the typos, small phone screen and fat fingers!)

TFortunato··on CEO says iRobot will never sell your data
Replied to a child post of this already, but in case you missed, they specifically said they won't share data either without first obtaining informed consent. (Full disclosure - former iRobot engineer)

https://homesupport.irobot.com/app/answers/detail/a_id/964/~...

TFortunato··on CEO says iRobot will never sell your data
They do say they will never share with 3rd party without consent, and will delete any data collected on request.

https://homesupport.irobot.com/app/answers/detail/a_id/964/~...

Full disclosure: I am a former iRobot Engineer

TFortunato··on CEO says iRobot will never sell your data
Hey, w/o getting too specific, speaking as a former iRobot engineer when the 900s were developed / released (unofficial, not representing iRobot, blah blah ) - the privacy implications were thought of a lot when these products were developed. There really aren't any images or data that could be reconstructed / assembled into images ever sent to the cloud, persisted on the device, or available to send to the cloud.

I do understand the pessimism, especially with all the privacy horror stories you hear out there, but I can say that Colin / iRobot arent playing games with their words when it comes to this.

TFortunato··on Luna – Visual and textual functional programming language
Thanks that sounds great! I just signed up for the mailing list, but feel free to reach out to me at any time! (my username @ gmail.com)
TFortunato··on Luna – Visual and textual functional programming language
yep, and I've used Labview (and Simulink, and other graphical editors) in the past for some of this.

Why I'm excited about this is: 1) It's open source, so we can extend it and hack on it as needed. Thos is probably the biggest reason. 2) The dual textual / graphical representation is really useful in cases where you might want to switch between the two, or one makes more sense than the other. You can do some with text in some the other tools, but the last time I tried, it felt like an after thought more than something expressly designed into the language. 3) It looks to be designed as a general programming language, focussing on having a good compiler, good tooling and a nice foreign function interface. My hope is that this means we can incorporate it into our existing code bases easier than labview, where it kind of wants to stand alone. As an example, perhaps I have a large existing codebase for my robot written in C++ with ROS. Perhaps I could use this to create a new ROS node that does my vision pipeline, by taking advantage of the fact it compiles down to native code, and has a foreign function interface designed in from the start. (Or maybe I'm dreaming, but my hope is it would be easier and more performant than doing the same in Labview!) 4) Did I mention open source :-)

TFortunato··on Luna – Visual and textual functional programming language
Super super cool project! As someone who does robotics, I'm super eager to try this, and to hopefully contribute to some open-source libraries for this.

A lot of work in robotics involves software that maps to this style well (functional-ish, where data is being pushed through computational pipelines / graphs), and I think this could be a killer development environment for things like control systems, sensor fusion software, image processing / computer vision, etc.

The fact that it is going to be open source, and already seems to have some nice support for things foreign libraries, profiling support, and well as visualizing of your data visually (that image processing graph example!) makes me think you are going to get a good response to this. I also think there are a lot of hobbyist type projects (RasPi level 'smart home' stuff, algorithmic art / music, SDR, anything you see on Make / Hack-A-Day) who would love a tool like this!

I'm very interested in checking this out, and in contributing packages / libraries if that will be supported. Hoping to get access to the alpha!

TFortunato··on American Chipmakers Had a Toxic Problem, Then They Outsourced It
Another fun Silane fact for you - The auto-ignition temperature is right around room temperature / a bit lower, and pure silane burns with "invisible" colorless flames!

I worked at a semiconductor equipment manufacturer, and the silane gas delivery systems had to have separate flame detectors (from the normal one's we used) in the gas boxes for cutting off the gas supply because of this risk. We also had special strobe lights in the building with a sign below them indicating "Silane Pad on Fire if flashing", to let us know the gas bottle out behind the building (gas was piped in) was likely burning, since you wouldn't know from looking out the door / window unless you got close enough to feel the heat!

Luckily, I never experienced a silane fire, but for a point, I did design and implement the safety system for an experimental ion implant system, which used silane along with other nasty process gases containing arsenic, boron and other dopants. Cool job, but the stress of "make sure you get this right" was a lot higher than other software jobs I've had, haha!

TFortunato··on Relational Reasoning with Neural Networks
Oh that is awesome, and I'll have to check the paper out! That's actually something I've been wondering -- about combining the power of deep learning with the domain expertise of old style expert systems, whether for bootstrapping the neural net, or as a rule base it can consult.

It seems like it would be very powerful to be able to harness Neural Nets for dealing with fuzzy inputs (images, natural language, spoken words), but adding a propositional rule-base they can consult for whatever the actual task is once you have dealt with that input.

On that, if it was learning a rule base, it might also be really helpful with getting insight into what your model is doing, if you could somehow introspect on / view the rules that it learned on a higher level than "when these neurons fire, we do this to the output"

(I woouldn't be surprised of course, to find one day that the DeepMind guys are already on top of it, and come up with a "Neural Warren Abstract Machine" at some point)

TFortunato··on Before Silicon Valley, New Jersey Reigned As Nation's Center Of Innovation
Yep...seems most of what I hear when people from Massachusetts (myself included) when we talk about where we work to outsiders is either "In Boston", or "Right Outside Boston / Near Boston / The Boston Area" (For anything east of Worcester it seems, haha)
TFortunato··on Relational Reasoning with Neural Networks
Very cool! Not quite the same, but brings to mind SHRDLU, which was recently discussed here.

The first few thoughts I had on seeing this: 1) Of course this is by DeepMind! Why would I think anything different. (I love the "basic" research they are doing on NNs & Deep Learning, and am always excited to see a new paper by them).

2) I would love to see more investment into this kind of basic ML research. (By that I don't mean "easy", but addressing the fundamentals of how to approach different types / classes of problems). A lot of where the DeepMind guys seem to be finding these big wins is in combining "classic" AI / CS techniques with Deep Learning / Optimization.

Examples (And I'm a novice at deep learning, so someone PLEASE PLEASE correct me if I'm wrong): AlphaGo - Take a technique like Tree Search for playing a game, and combine with deep networks for the tricky bit of evaluating play positions Deep Reinforcement Learning - Q-Learning and other reinforcement techniques have been around for a while, but they adapted them to a deep neural net architecture Neural Turing Machines - Took a classical model of computation and made it differentiable, alowing for a neural net to "learn" algorithms like sorting. Deep Neural Computing - Figured out how to add and address external memory in a differentiable computer, allowing a neural net to solve problems like path finding on a graph.

Where I think a lot of cool stuff is going to continue to come from is by revisiting classic techniques, and figuring out how they can be adapted to a differentiable / optimizible architecture. Or taking a classic problem and finding an efficient way to evaluate "goodness" of an answer that lends itself to being used in an optimization problem. Again, not saying it is easy, but I wonder how much "low hanging fruit" there is in revisiting classic algorithms and GOFAI techniques, and asking "can I use this in a Neural Net or adapt this to be differentiable so that I can learn or optimize the tricky bits?"

I'm sure I'm glossing over a lot / missing the point of a lot of it - like I said, just a noob whose super excited about this stuff :-)

TFortunato··on Human-Level AI Is Right Around the Corner – Or Hundreds of Years Away
Perspectives from some well known names here at HN (with special shout-out to my former, expert level shade throwing boss, Rod Brooks!) as part of a larger report: "Can We Copy The Brain" http://spectrum.ieee.org/static/special-report-can-we-copy-t...
TFortunato··on Realtime Bitcoin
Waiting for the inevitable: "What if we could guarantee that these coins were made only with renewable / sustainable energy sources, giving you the benefit of digital currency w/o the pollution", and the coming "It's like CO2 / renewable energy credits, but with blockchain!" startups.
TFortunato··on Revolution Pi – Industrial PC Based on Raspberry Pi
Unfortunately, as awesome as that is, that pretty much means industrial "in name only"...at least for non-hobby / non-startup projects
TFortunato··on Sand and gravel mining “greatly exceeds natural renewal rates”
Did you even read the story... it's not a "sky is falling" story. (A Miami is falling story, maybe :-P)

It's really more of an interesting look at a resource that few in our part of the tech world even think about, and some of the challenges they face, if you can look past the headline.

TFortunato··on A crashed advertisement reveals logs of a facial recognition system
I tried the demo. A little sad, that it sees me as 12 years older than I am, and that I apparently always look angry and disgusted! I'm going to blame it on my glasses and bushy beard, and try to look at the bright side - apparently face scanning systems aren't quite good enough to get a read on me yet. (And try not to be too sad about looking like a grumpy old man)

(Anyone else with a glasses, a beard, or other non-typical facial features want to comment? I'm curious now how well their system handles these?)

TFortunato··on A Single Autonomous Car Has a Huge Impact on Alleviating Traffic
Yeah, not sure what the big news is here... "traffic waves" actually have been talked about for a long time... I remember reading "the science hobbyist" pages almost 20 years ago (!), about this. I think besides the fact that the car is autonomous, and maybe more consistent than a human driver would be in this situation, there is nothing really new here.

http://trafficwaves.org/

TFortunato··on ODrive Project – Using inexpensive brushless motors in high performance robotics
Very cool! So I should perhaps qualify my comment to say that some newer appliances may be moving that way. (I've done work on motor drives for consumer electronics, robots and some industrial devices in the semiconductor world...no appliances, but it's cool to hear some of them are going that way!)
TFortunato··on ODrive Project – Using inexpensive brushless motors in high performance robotics
Pretty much this! People definitely do work some black magic here. (See for a simple example microstepping drivers, where the current profile isn't sinusoidal, but specfically tuned to match the motor profile).

One place where I did see this used frequently is in certain consumer electronics, such as inkjet printers. (Ever wonder why companies like Allegro sell weird combos like a chip for driving 2 steppers, a DC motor and a Buck converter....this is why)...When you are shipping a million printers a year, you save pennies using the cheap actuator and fix it in the controls, but for most projects, yeah totally not worth the hassle!

TFortunato··on ODrive Project – Using inexpensive brushless motors in high performance robotics
Hmmm, not sure I follow you there, care to explain?

(Unless you are referring to block commutation of the BLDC phases, which can look sort of similar to step wise control, but I would rebut that in precision servos drives, we would also look at doing non stepwise commutation (sinusoidal, FOC, etc.))

TFortunato··on ODrive Project – Using inexpensive brushless motors in high performance robotics
Yeah, brushless is definitely the way to go for efficiency, and good smooth motion. Stepper for an e-bike or washing machine (Not a BLDC either, but...) would be nuts, and in general steppers are really for cheap semi-decent control, not for anything where precision is really needed.
← PreviousPage 7 of 9Next →