Sign up for our monthly newsletter so you never miss the latest from InDepth!
By Michael Menduno
I am a water person. I swim four to five days a week. I am a diver. I’m also a technologist and gearhead. So, when I saw the FORM ‘smart’ (augmented reality) swim goggles, I knew I had to try a pair.
FORM goggles offer the swimmer key user selectable metrics such as time elapsed; distance swam, split times, interval times, pace per 50/100 yards or meters, stroke rate/SWOLF, or even calories burned in a concise, bright-yellow heads up display that appears to float in your field of vision. Note that the metrics are based on the trailing length; they’re not instantaneous. The heads up display also cleverly sequences metrics as the swim progresses. For example, it shows the new split time after the swimmer performs their turn. Also, the eyepieces are symmetric so that display can be positioned in either your right or left eye.
As a result, you don’t have to stop swimming to look at your swim watch or deck clock for timing, and it counts your laps in case your early morning swim brain loses track. According to the company, the goggles free up your mental bandwidth so you can focus on your form, hence the name. Did I mention they currently have no competition?
FORM goggles can be readily set for lap swimming, intervals or drills (ah those kick sets!) with the two tiny toggle buttons on the side of the goggles; they don’t have GPS and so don’t work for open water swims. At the end of your workout, you save the swim using the menu buttons, and it’s uploaded to the companion app on your smartphone via Bluetooth, where it is stored and can be shared. Your swim data can also be uploaded to Strava or Training Peaks, though currently FORM is not integrated with Swim.com, which partners with the U.S. Masters Swimming (USMS) log. That’s where I upload my Apple watch swim data.
Oh, and the goggles also talk to the Polar OH1 optical heart rate sensor, which can be fitted to the google straps, allowing the device to display your heart rate in real time. They can also upload heart data from a Garmin watch. All in all, FORM can only be described as a remarkable piece of engineering.
Interestingly, when I first called the Vancouver B.C. Canada-based company to inquire about their goggles, one of my first questions was, “Are you also planning to offer an augmented reality diving mask?” NOT!
FORM’s marketing director explained to me that there are an estimated 30 million swimmers in the U.S. and 240 million globally making it the largest sport in the world—roughly one to two orders of magnitude larger than diving, including both free and compressed gas divers—and that’s who they plan to focus on. That of course would include triathletes and Iron-people who swim too.
I also realize now that FORM’s secret sauce is a lot more than simply displaying computer data on an in-mask optics display. Though the diminutive optics screen, which is about the size of a 64GB SanDisk card, and the small thumb drive-sized computer positioned at the temples, could easily be incorporated in a diving mask. Shearwater and Thalatoo take note.
The Ghost In The Machine
FORM is the brainchild of 44-year old competitor swimmer cum mechanical engineer, Dan Eisenhardt. He first conceived of the idea for smart swim goggles for his MBA project in 2006, but the available technology, particularly the state of machine learning (ML), at that time made it infeasible. Instead, he and his colleagues created a heads-up display ski mask and went on to form a venture-funded start-up called Recon Instruments, which Eisenhardt incorporated in 2008. He eventually sold the company to Intel in 2015, after creating five generations of smart glass products. He left Intel in late 2016 to return to his original vision, creating a smart swim goggle.
Compared to a swim watch, which is attached to a single ‘stroking’ appendage; detecting starts and stops and strokes—not the Irvine kind (inside joke)—from a device positioned on your head is arguably a harder problem. But FORM appears to handle it with aplomb. Eisenhardt told me that they want to be accurate to within 99% of the metrics and it’s rare to see a mistake. Interestingly the goggles were able to correctly detect my stroke (fly), when my Apple watch didn’t. More on that later.
At the heart of the goggles, is a three-axis accelerometer and a three-axis gyro that interprets what’s going on in the rest of the body i.e. doing freestyle, doing fly, cranking out 80 or 115 strokes a minute, burning calories, based on that sensor input. That’s where machine learning comes in. FORM trained its algorithms using machine learning, with actual data feeds from many different types of swimmers, including the award-winning University of British Columbia’s Thunderbirds swim team, who were all recorded with video cameras. The various types of movements were labeled and then fed back into the algorithms, which were then tweaked and tested again, etc. Rinse and repeat!
Eisenhardt compared machine learning to opening up a Pandora’s box. “We had no idea back then how hard it would be,” he said. “We would change something over here and it would get better, but then something over there would break. It was always a moving target so we had to find an optimum point based on our goals and objectives. That was a big one!”
Other challenges? “It ends up being a combination of things, a kind of an engineering matrix,” Eisenhardt explained. “Low-power optics, machine learning, and bulk; those were three massive constraints, above and beyond price. Because price, is actually the most important. You have to constantly have that price rationale in the background.”
FORM goggles retail for US$199. By comparison, ordinary swim goggles sell for US$15-30 up to US$59 for my MAGIC5 custom-fit goggles. Of course, like many of my swimmates, I have accumulated quite a collection of goggles over time.
I found downloading and setting up the FORM app and goggles was easy. The goggles are high quality with a soft sleeve and come with a selection of different sized nose bridges for an optimal fit. I found them very comfortable and watertight, and they didn’t fog. Though they weigh about twice that of an ordinary pair of swim goggles: 2.3 ounces/65 grams compared to about 1.3-1.5 oz/37-43 g, I didn’t notice any difference once they were on my face.
My first experience using the goggles in an outside pool was disappointing. While the heads up display and metrics, which I had positioned in my right eye, were brilliant, I had a hard time seeing out of my left eye. FORM’s lenses are barrel-shaped with a flat forward surface. For me, it looks like a sunglass lens in the middle of my field of view. What’s more is that I kept seeing what looked like cavitation bubbles along the flat surface. As a result, I felt like I couldn’t see the pool well. My view also looked a bit wonky when I approached the wall to turn. FORM says the goggles take a little getting used to. Yup! I tested them in two different outdoor pools with the same result.
I asked Eisenhardt about seeing cavitation bubbles. He said that only two other people out of hundreds or more reported this problem, both of them in outside pools; something about the light perhaps. So I tested them again in an indoor pool. This was my fourth workout session with the goggles. The sunglass effect was definitely diminished, and though I still saw some cavitation bubbles against the glass, my visibility had improved, though not as good as my MAGIC5, which of course, does not include a heads up display. This made me think that with time my vision and or brain would adjust.
I mentioned discovering that the FORM goggles correctly identified my butterfly while my Apple Watch 4 did not. This is often a problem with the watch. It also has trouble identifying breaststroke. Even more interesting, when I compared my split times from my watch on swim.com to those of FORM (I was wearing devices during my trials), the times showed significant differences sometimes amounting to a few seconds. A little unnerving! How fast do I really swim?
This, of course, raises the question, “How accurate is the [name of device]?” The answer of course is, “Compared to what?” It would be useful to compare both to an electronic pad used in swimming comps.
I was already aware of timing problems with the watch. For example, ending a lap by touching with my [left] watch hand yields a different time than touching with my right. Eisenhardt said that FORM’s accuracy is based on matching swim data to video recordings to within a certain tolerance, which is how they derive their 99% accuracy figure. “It’s very rare that you would touch the wall and get a time where you’re like, oh, that doesn’t seem like the right time. It’s just a very rare event. It’s hard to say exactly how accurate we are, but we are definitely accurate enough for you to never have to second-guess a metric.”
So there you have it. I LOVE the FORM display and the data; however, I am not yet happy with the visibility. I wish it integrated with the Swim.com cloud and consequently my official USMS log. ;-( I am also now intrigued with interval and split times, and plan to do more investigation with both devices. Are the FORM goggles worth $199? Never mind how much I’ve already invested in swim goggles, does the money really matter for a sport one feels passionate about? I am going to keep swimming with FORM, and keep my watch on, for now.
Heads up people. Watch this space.
Additional Resources (from FORM)
Michael Menduno is InDepth’s executive editor and, an award-winning reporter and technologist who has written about diving and diving technology for 30 years. He coined the term “technical diving.” His magazine “aquaCORPS: The Journal for Technical Diving”(1990-1996), helped usher tech diving into mainstream sports diving. He also produced the first Tek, EUROTek, and ASIATek conferences, and organized Rebreather Forums 1.0 and 2.0. Michael received the OZTEKMedia Excellence Award in 2011, the EUROTek Lifetime Achievement Award in 2012, and the TEKDive USA Media Award in 2018.
They Helped Foment a Dive Computing Revolution: RIP Cochran Undersea Technology (1986-2020)
Latin American distributor for Cochran Undersea Technology and certifiable dive geek Carlos Lander recounts the many firsts and innovations in dive computing created by microprocessor pioneer Mike J. Cochran (1941-2018). These included the first wireless air-integrated dive computer, automated sensors, hands free gas switching, tap interface, compartment-level conservation factors, what-if software, and the “Cochran Navy”—used to by the US Navy, to run their proprietary VVAL 18 algorithm.
By Carlos E. Lander
Header image courtesy of C. Lander
Carlos Lander was the distributor for Latin America for Cochran Undersea Technology.
Michael James Cochran, genius and founder of Cochran Undersea Technology, revolutionized the design of dive computers (DCs) with the company’s state of the art US Navy Computer that impacted the entire DC industry. But before exploring the accomplishments of Cochran Undersea Technology, let’s answer an integral question: “Who was Michael James Cochran?”
Mike was born in Daytona Beach, Florida, on May 21, 1941. He worked on a missile tracking ship as a young adult but also had an illustrious career in electronics, receiving 57 patents, including ones for the microprocessor and microcomputer chip.1 The microcomputer chip patent was assigned to Texas Instruments (TI)—where Mike worked for many years—and issued to Gary Boon and Michael J. Cochran in July 1971.
For his work on that microcomputer chip, Mike won an IR-100 Award from Industrial Research Magazine (now known as R&D World Magazine). After he left TI, he worked with NASA and invented a DC for their astronaut training program. He founded Cochran Undersea Technology in 1986, and began designing and manufacturing DCs for recreational, sport, commercial, and military diving applications.
In 2016, he was awarded the honorary position of Admiral in the Texas Navy, an accolade commending exceptional community service. In that same year, he won the New Orleans Grand Isle (NOGI) Award—frequently known as the Academy Award of Diving—which recognizes pioneers of the underwater world.
Sadly, Mike passed away on December 2, 2018, at the age of 77, leaving behind a lasting legacy of studies and research in electronics. Mike was always driven by passion—so much so that he worked until the end of his life, having never created a plan for Cochran Undersea Technology to continue without him.
The Cochran Undersea Technology employed brilliant and gifted people such as Martin Heerschap, designer and engineer for the Cochran closed circuit rebreather (CCR) and liaison to the US Navy throughout the Navy DC’s development. Sales manager and tech instructor Larry Elsevier was another who worked closely with Cochran clients including the US Navy and NATO. He passed away in 2014. There was also Jeff Loudan, physicist, mathematician, and software engineer; Stuart McNair, engineer; and John Corso, talented diver and the face of the company before Mike’s passing.
Creating An Advanced Dive Computer
Everything started in 1991 when Cochran Consulting Inc—the parent company to Cochran Undersea Technology—filed a patent for an “advanced dive computer,” which was intended to become an Oceanic air integrated DC.2 While that specific product was never patented, Cochran’s DC led to a huge advancement in the field. First, Mike made a DC from a Single Board Computer (SBC), and then he programmed everything in an “assembly language,” securing very high-speed, real-time calculations on a reliable DC.
An SBC refers to any computer that contains all of its components in one circuit board. This configuration is perfect for devices with limited hardware space, such as a DC. SBCs are self-contained and energy-efficient, important elements under diving conditions.
An assembly language is an architecture-specific, low-level programming language, and Mike used one to efficiently compile a very complex algorithm including a set of variables that, at the time, did not exist in a DC (breathing parameters being one).
The final product included a microprocessor, a depth and pressure transducer, an electrically conductive metal clasp, and other components that worked efficiently in a small package within a very low power consumption unit.
Consequently, Mike decided to produce his own line of DCs, dubbed Nemesis in the United States, and Aquanaut in the EU. Although the company worked with different brands, in the end they decided to concentrate on their own, the Cochran Dive Computer. No other manufacturer at the time designed and built every component of their DC in-house.
Cochran was the first company to produce a hoseless DC with the following design features:
- Mike’s design allows the main computer to be attached to the tank, so all the information gathered by the pressure transducer is transmitted to a diver’s wrist monitor (a second computer) in real time3, including the diver’s breathing parameters and workload. The wrist unit could be used independently as a DC in the event of a tank unit failure. Other brands used a transmitter attached to the tank, limiting the amount of information that could be sent to the main unit in short data bursts.
- As determined by Cochran, the cases on both units were air-filled (at 1 atm), ensuring that the cases were manufactured with material that was capable of withstanding extreme depths. Taking this approach required fewer case penetrations—such as buttons—and ensured an effective, long-term seal. In addition, the battery compartment was sealed from the electronics and built with materials that wouldn’t corrode. In an extremely rare case of flooding, the only damage would be to the battery compartment and not the electronics.
- The case was equipped with three electrically conductive metal clasps instead of pushbuttons. Those stainless contacts, in conjunction with the electronics, could detect the difference between saltwater and freshwater and thus refine the depth calculation. They could also distinguish metallic objects and fingers via electroconductivity.
- Another cool patented feature was the implementation of a vibration detector inside the unit which allowed the user to perform quick functions, like tapping the unit for five seconds to turn it on, or tapping it once to turn on the back light.
- The Cochran computer accurately measured and recorded the altitude (pressure) every minute whether it was on or off, accounting for minute changes of nitrogen levels in tissue.4
- Other contemporary DCs accounted for changes in decompression conservatism, and while some commercially available computers offered conservatism customizations, they didn’t provide adjustment calculation guidance. The majority of DCs based the conservatism factor in changing altitude; instead, Cochran used gas-loading to add conservatism in proportional increments so that divers both understood and controlled their conservatism.
- Cochran’s DC’s tandem PC software featured two capabilities not offered by any other manufacturer at the time. First, the PC software ran the same algorithm as the DC, behaving exactly as the DC did on a dive. Divers could perform a trial run on the PC graphically or review an existing profile. Secondly, divers could transfer gas loading and altitude characteristics from one DC unit into another DC unit. For example, if one computer malfunctioned, information would transfer from one unit to another. Therefore, divers could continue to dive with the new computer without noticing any difference.
- Perhaps most importantly, the Cochran DC never locked a diver out; most DCs would block access for 24 hours if the diver violated a stop. Cochran’s DC allowed them to continue their dive, and the DC would continue off-gassing at their current depth.
The Proprietary Cochran algorithm
Mike exchanged ideas with dive experts while developing Cochran’s proprietary algorithm, including Dr. Bill Stone, Dr. RW Bill Hamilton, and Capt. Edward Thalmann of the US Navy. He designed the circuit board with the algorithm in mind. During this period, Capt. Thalmann was developing the Navy’s proprietary VVAL18 algorithm with the intention of using it in a forthcoming Navy DC. His plan was to be able to test the DC on the manned test-dive data to evaluate its performance against the algorithm.
Once Cochran Undersea Technology was awarded the contract to build the Navy DC, they had access to all the scientific studies, probabilistic software, and anecdotal data on the installation of VVAL18 into a DC, which was pivotal to the development of Cochran’s own decompression algorithm. Cochran’s algorithm for the Nemesis included compensation for a bubble formation, and the Gemini—based on the Nemesis prototype—added variables for ascent rate and breathing parameters. The algorithm is based on a modification of Haldane’s decompression model, with compartments between five and 480 minutes.
There were several versions of the Cochran algorithm: 14, 16, and 20 compartment models. In the case of the 20 compartment version, the algorithm included fast compartments to compensate for helium gas and added a compensation for microbubbles related to ascent rate velocity. The model also used the same linear off-gassing from the Thalmann algorithm but included more than the fast compartment and ascent velocity to compensate for the aforementioned microbubble formations.
DC algorithm evaluation has been the subject of some research. For example, Dr. Carl Edmonds compared DC responses to a series of bounce dives. Dr. Karl E. Huggins used the same technique to evaluate DC algorithms by testing them on profiles that had known human subject results.
In his article published in 2004, Huggins notes that DC manufacturers did not validate their algorithms with human subject tests, so running a DC against a battery of previously tested dive profiles provided some rudimentary level of validation. So, when it came to validating their algorithm for use in both the Nemesis and Navy DCs, Cochran had the advantage of accessing the Navy’s database of man-tested dives.
Along with validation against the Navy’s database, the Cochran DC was validated against NOAA’s custom tables on the wreck of the USS Monitor, and the tables and DC were said to have matched well throughout the project.
In a 1989 Undersea and Hyperbaric Medical Society (UHMS) Workshop on Validation of Decompression Tables, UHMS determined that decompression algorithm validity could only be proven using primary data, such as results derived from controlled laboratory conditions. However, secondary data such as anecdotal performance reports could be cited as an operational evaluation but wouldn’t be considered proof of validity.
Cochran’s Navy DC and what made it so special
In 1996, there were no commercial DCs running the VVAL18 decompression algorithm. The US Navy Experimental Diving Unit (NEDU) sought a manufacturer to install the VVAL18 on a DC following their specifications, which were far from simple.
Cochran won the bid (being the only DC designer and manufacturer entirely in-house likely had much to do with it) and delivered five modified Commander DCs with the VVAL18 algorithm, which they called “Cochran Navy.” After a few modifications, Cochran developed additions to the Single Board Computer (SBC) that allowed for massive dive profiling memory and a one-second sample diving profile (the Navy required a maximum of two seconds). The finished product was a DC that handled large amounts of data and self-test diagnostics.
Due to the breathing parameters, the DC would operate with air if the depths were shallower than 23 m/75 ft, or at PPO2 = 0.7 (MK16 MOD 0 UBA) at further depths. Additionally, hands-free Gas Switching was implemented to eliminate the need for buttons5. Part of the computational power was required for this, as the switch function would depend on depth and time. Also, the DC was programmable on the surface or via PC.
The device computed decompression properly whether the diver was below or above the stipulated stop. The residual gas was based on the diver’s depth: a real-time-calculator, without gimmicks. Therefore, the DC never shut down or left the diver hanging. Other advances were handling of the magnetic signature, EMF emissions, and visible light emissions (red light) required for Explosive Ordnance Disposal (EOD) work, and for stealth. The DC could be programmed for the needs of the mission with the Navy’s Analyst computer software.
According to probabilistic decompression models for the profiles tested on the Cochran Navy DC, the average risk predicted of decompression sickness occurrence was low: less than 1%, as expected. Therefore, the DCs were validated by faithful replication of the decompression schedules when exposed to simulated manned tested-dives.
The computers made by Cochran Undersea Technology, while advanced for their time, were misunderstood and misused by many. Still, I consider Cochran’s Gemini DC the best ever made. If you’re not breathing from the back-gas and can reach certain pre-programmed depths, the computer automatically knows that you’re breathing from the deco bottle. If you start breathing from the back-gas again, automatically switching back without needing to push any buttons is an astounding characteristic unmatched by any other DC.
Unfortunately, the absence of a marketing division in the company gave Cochran the reputation of producing a military-only DC. The EMC-20H, a later DC, was also very advanced for its time, so it was sad to see Cochran vanish. I am comforted by the fact that I still have a few Cochran DCs I can use, and that they will serve me for many years to come.
I want to thank Martin and John for our conversations, and for their help in getting the facts straight.
- Patent No. 4074351
- Patent Nos. US4949072, 4999606
- Transmitting information every second reduces noise; the power transmitter and a 250 kilohertz frequency results in a strong communication between units and protection from interference with other devices such as camera strobes and scooter motors.
- When a diver changes from a lower altitude to a higher one, the computer detects this change and adds nitrogen to the tissue compartments. The differential pressure between the nitrogen in the body and the higher altitude must be equalized (outgassed). Conversely, when a diver changes from a higher altitude to a lower one, the computer detects this change and removes nitrogen from the tissue. The computer automatically reacts to long-term stays at a constant altitude. If a dive is made while at altitude (whether the computer has already automatically reacted or not), the nitrogen algorithm within the Dive Computer is adjusted depending on the exact depth.
- Patent No. 5794616
Cochran Undersea Technology Technical Papers:
Carlos Lander—I’m a father, a husband, and a diver. I’m a self-taught amateur archaeologist, programmer, and statistician. I think that the amateur has a different mind set than the professional and that this mindset can provide an advantage in the field. I studied economics at university. My website is Dive Immersion. You can sign up for my newsletter here.
Thank You to Our Sponsors
Why Do Divers Run Out Of Gas?
Not surprising, the answer is more complicated than simply, they neglected to look at their gauges. Here Aussie diving medical...