I recently had the chance to see one of Cobalt's robots on patrol at night in an office in San Francisco. It looks quite slick, like how Apple might have designed it. Well done.
The fixed hardware costs and limited capabilities have long kept robotics in the hands of academia and manufacturing. As those cost curves move, we're now seeing viable business models in non-industrial B2B robotics, like this one. I expect the self-driving car growth will also push down the costs of LIDARs and other critical enabling technologies that will eventually lead us toward practical consumer robotics.
Hacking these will be the new frontier in industrial espionage. Imagine the possibilities: listen to employee conversations, video access to sensitive areas including computer screens, property damage, personal injury...
Yeah if anything this takes the attack surface and multiplies it quite a bit. Imagine if one could sneak something into a firmware update for all the Cobalt bots.
First of all, very nice engineering in here. Well designed, appealing look, functional and has a good feature set.
However, how is this bot better than current CCTV with some checkpoints, badge readers and all the same features that we current have? A CCTV guarantees coverage by design, while this robot has does sweeps that allows intruders and all to evade detection.
Of course, we have security guards walking around buildings with checkpoints, but they are usually added to CCTV, because people walking around is more of a deterrent than a recording.
I saw in the end of the article the idea that it would be less expensive than a CCTV, but they do not give a price point and I really doubt. Even high-quality colored cameras with infrared sensitivity are cheap as hell nowadays. Most of the cost is in the back-end services (monitoring and recording), and unless the company is working on improving this (the text does not mention), I don't see how they would compete.
These is a sincere question - the company is obviously doing at least somewhat well and it's not my intention to be the next BrandonM. As a hardware engineer, I would even work for them, as the project seems really cool and there are lots of room for new ideas. But it looks very expensive and this is a crowded, cost-oriented market.
Edit: Also, is it common in the US (I'm Brazilian) for security personnel to walk inside offices while people are working, as depicted in this video? Here I only see this during off-hours.
I was initially skeptical... but Cobalt was a problem-centric creation. We started the company after speaking with security practitioners who were clamoring for after-hours coverage & insisted that cameras & access control just couldn't provide the responsiveness & customer service that their organizations demanded. CCTV cameras are great for entry and egress. But they suffer from a host of issues: extremely costly to install, upgrade, maintain, and monitor (upto $75/mo/stream); are seldomly upgraded (analog cameras are still common!); huge bandwidth consumption to cover large areas; provide no real-time responsiveness; and are only used after the fact for investigations. But guards were just so expensive -- so companies just left major gaps in their security coverage.
We see the robots as a compliment to existing static infrastructure, where the systems automatically communicate & work together. In fact, some of these systems are built into the robots too: cameras (obviously), access control badge reader, and a heck of a lot of compute & other sensors. The concept is quite simple: Mobility is becoming increasingly cheap, and enables you to reposition your most-capable resources to the right place at right time. It's the most capable edge sensing & computation feasible, and the human-in-the-loop lets you provide a host of other capabilities too. In effect, we're allowing a single, remote person to be in many locations at once -- particularly during the low-utilization after hours (nights, weekends, & holidays). It's similar to the force multiplication that CCTV provided decades ago!
I am completely failing to see the "security" (as in "security guard").
It seems a clearly smart way to "keep an eye" over an area, but more like in "surveillance" than "security".
I mean - pardon me if I am still skeptical:
1) how would the robot deal with (locked) doors (or even with doors that need to be opened through a pushbar or handle)?
2) What happens in a multi-storey facility? (I mean can it go up and down stairs or use a normal lift)?
3) Is the communication protocol between the robot and the control unjammable and uninterceptable?
4) How - besides sending an alarm to a human-presided control center with an intervention time (I presume) in the order of magnitude of several minutes - can it deal with an intruder?
Night & weekend guards (almost universally!) do not carry firearms or even non-lethal weapons. Their job is exclusively "observe and report", and they are not permitted to physically intervene -- as a matter of policy to limit liability.
>Night & weekend guards (almost universally!) do not carry firearms or even non-lethal weapons. Their job is exclusively "observe and report", and they are not permitted to physically intervene -- as a matter of policy to limit liability.
Yep, and those are generally called not "security guards", but rather "night watchmen" or "night guards".
Congrats! I love how Cobalt's robots are super friendly-looking, and also how they can make big, empty buildings feel safer and less scary for people who have to work nights or weekends. I've definitely been nervous working in those situations, and would have loved to have a Cobalt robot as a comforting presence nearby.
The fixed hardware costs and limited capabilities have long kept robotics in the hands of academia and manufacturing. As those cost curves move, we're now seeing viable business models in non-industrial B2B robotics, like this one. I expect the self-driving car growth will also push down the costs of LIDARs and other critical enabling technologies that will eventually lead us toward practical consumer robotics.