Ottava Surgical Robot : First Early review and sadly concerns
- Steve Bell
- Jun 22, 2024
- 18 min read
Updated: Jul 30
Okay - take this as what it is. It's knowledge of surgical robotics, it's patent scraping, it's early image analysis, and above all, specultion on those initial images.
But here's why I'm concerned for Ottava from Johnson & Johnsion Medtech.
Note this may be completely wrong but in lack of other evidence - these are my conclusions

So the concept of Ottava surgical robot is a bed that forms the body of the robot - and then 4 arms that come from under the bed - come up past the patient and reach the patient.
Basically the bed is the robot.
The concept is that it has "zero" footprint in the OR.
It has integrated table motion "Twin Motion"
The arms fold away under the table when not used so it "saves space"
So today I got the first non patent images (Credit Tom Salemi) of the Ottava... and well.. I've been concerned for a while... and now I'm super concerned.

This is an above view down to the Ottava. What you can see is the bed - with split legs - and it looks like s standard bed dimension. Coming from under each corner of the bed are 4 arms of the Ottava robot. They tuck under the bed when not in use - and with the push of a button they pop out ready to use.
I can already see several issues with this design. (and they do look very Versius by the way.)
The concept
The concept of a bed mounted (or rather bed contained) set of robotic arms is not new at all.
It was conceived back in the late 1990s as Zeus and several other designs and prototypes.
I personally have been involved with robots that were to be bed mounted.


The idea back then was to make a "zero footprint" and an integrated table motion. And well... they failed. And for good reasons.
Bed mounted is an issue
If you look at the above image of Ottava - a keen eye - and people that have supported hundreds of OR cases will note a few things. The Ottava arms look nice and slim. But for them to fit under a standard bed there are several compromises engineers must make. One is potentially joint lenghth. There is a finite amount of arm length you can make betwen joints if you want them to fold and fit under the bed. And that then limits the reach of the arms.
With simulator boxes that have simplified and defined trocar placements. Or with cadavers - you can feel you make the arms reach at all times. But that is seldom representative of the real world. Controlled environments don't reflect what you will really find out in the real world.
If there's one thing I learned in attending hundreds of robotic cases - you need way more distance and height differentials than a fixed to the bed arm will potentailly allow. More often than not, cart based arms, or da Vinci booms need to be at the limits of joints. Fixing the arms to the bed puts major constraints on arm positioning - and that can lead to clash issues or worse- inability to even set up the case properly due to lack of reach or terrible arm poses at joint limits.
It is not just in the X and Y access that arms can struggle to reach target anatomy - but of more concern is how that height adjustment in the Z axis can be reached. The top image indicated to me that these arms will struggle to find good poses at extremes of body habitus- or big domed abdomens. There just will not be the variety of angles needed to get to good positions to make use of the best kinematics. And that might refelct in issues of port placement.
I'm predicting that for many patients the trocar positions will end up being compromised and scope angles- instrument angles will be an issue. This will potentially cause arm phantom clash, and even full clash that could end up in alarms.
Workflow with Ottava surgical robot
Beyond positioning of the arms in a standard procedure - bed mounted arms create a work flow issue. The main issue is that the arms cannot be easily draped prior to the patient being brought to the table. (there may be a tuck away fold position when draped - but if that is below the knee line it will not be acceptable in many countries.)
So that would lead to patient induction - positioning - draping prior to arm draping. And most other systems can be draped during room set up while the patient is being induced. So this could lead to work flow disruption.
If they've got that worked out - the simple workflow of the robot bed is odd - as the bed is linked to the robot. That inherently means the robot is tied up for the entire induction - perioperative phase and ... post robotic usage. The robot is tied up until the patient is off the table and gone!!!
Most other systems in high utility settings are being stripped down, cleaned, moved and draped up in the next OR - while the first patient is being closed, recovered. This is a potential issue for genuine workflow. The locking of the arms to bed means that the bed and robot cannot be moved independently. So you are effectively locked tot he patient all the time.
I think that this design may work for USA anesthesia protocol - but I'm struggling to see how it works with various anaesthesia protocols and induction protocols around the world. If for example there is an anesthesis room - where they normally use a system of twin tables and pat slides... how will that work? Often induction is happening (epidurtals etc) in the anaesthesisa area ahead of the case. It will start to be logistics issue?
Thoracic & TORS with Ottava
The fact you have north south bed polarity of arms makes me wonder (but here I could be miles off) how you would use this arm set up in say a Thoracic case. I'm struggling to understad how you can bing three of the arms to the upper body segment in all thoracic cases. You might do some but not all. How will you be able to get this set up for all RATS cases. Or maybe I'm missing something? It may be way more flexible on the J1 start position than I can see.
It is absolutely not feasible to do TORS (Trans oral robotic surgery) with this design n my opinion. No TORS means exclusion from many hospitals and a high proportion of EU tenders. You need arm clustering and extesnive hight above the table to be able to do TORS. It seems impossible with this under table design. You just can't get enough height. Someone correct me please.
Zero footprint of Ottava
One advantage of this system is that it has no bedside unti on the floor. The bed is the only footprint. That is good in freeing up some space around the bed - but it does not automatically translate into bedside freedom of arms. Only a small part of the issue is the footprint on the floor (in modular systems). Arm real estate at bedisde is equally important. But a da Vinci Xi will free up the entire side of the bed - working space and access is just as important as footprint at bedside.
This design should help. But isn't what people may think.
Surgical Bed differences worldwide
Something tells me this has been a primarily USA base usability analysis for this system. There are plenty of hopsitals around the world that have moved away from this type of simplistic OR bed and have moved to say "column" based systems. Where there is just a transfer of the top part of the tabel. How will this work?
There is also a reaon why there are hundreds of bed types in the OR. But my biggest concern is what if the surgeon needs a bigger bariatric size table, or a more specialist table? Tables around the world are quite specific - will one size fit all? Maybe.
Many ORs have integrated table movement into the smart OR - will this table work with that? Does it have the full range of adjustmets needed for all specialties? Or do the arms limiteangles of table motion? It just feels odd and a possible point of resistance to force a surgeon and OR staff to adopt a new table.. just for those cases..
Oh and not let us forget that beds are not ususaly that dry. Do we really want fluids running down into the complex body of a robot? Think of big open cases with lots of washout. That bed is going to need to stay clean around those arms.
Clash and Z rails

One of the main issues you contend with in a robotic case is arm clash. And to avoid it you often need things like the 8cm rules on trocars. ie minimum gaps betwen trocars - trocars - or other structures,. With 4 corner constrained arms - I do wonder if to get clash avoidance you will end up with a dreaded 8cm rule? ie the distance between port and other ports or structures needs to be 8cm minimum in every case. Not so normal in laparoscopy and very annoying in small patients. And it means you need to learn new rules for port placement.
Twin Motion for Ottava
This has me scratching my head. Any robot with arms fixed to the bed gets "twin motion" by default - such as MIRA - or the even the Iranian robot - Sina surgical.
In many systems (even though you can technically keep instruments inside the patient) the advice when moving the patient (rather changing head up - head down - roll right - roll left) by moving the bed is to actually undock the arms. Like others with integrated table motion, Ottava seems to claim they can keep the instrumets in place. Which is all good until the patient slips and those scissor near a major structure - like a vein, or the liver - now pops into it. You won't see this rare event until a few thousand cases in... but when it happens?!!!
And some people will say - "It never happens" - so if that's the case why do a lot of robotic users add in extra "anti slip" pads to patients? Just one major patient event from this and through caution - instruments will be recommended to be removed from then on - and that feature is gone.
The picture at the top shows clearly that the arms arms are fixed to the bed - not the patient - and that is a slight nuance that is different that say a MIRA (Plus Instrument types, the way it docks to the table and a few other factors.) I have less concerns for a MIRA.
And I know they claim that gravitaional adjustments through table motion is critical. In the middle of the case it might happen once or twice (it's not continuous). But for me in the back of my mind - this way of doing table motion is "nice" with 50% of surgeons in my poll saying they'd like it. But I don't see it as a killer feature (well it might end up as one.) As I see the design roll out - I'm having my doubts on real world applicability (and highly controlled cadaver labs do not count as evidence the patient won't move.)
So I'll be watching that one closely and maybe I'm wrong. I'm often wrong.
Patient access in emergency failures
Irrispective of the company. Irrispective of the design of the robot and all the testing - now and again for many reasons (real world) the robot will go into a fault condition. Medium or high prioroty alarm. And that is very often when you go to mass production (hard) and after any software or hardware updates. It's just the nature of the beast. The fault rate in labs with "wizard engineers" attached to the robot with lap tops - is just lower than happens when the system gets into the real world with no enginering support via laptop.
I've been in many cases with many robots where they have locked up - medium priority alarm on one arm or even system wide alarms.
So if you have a design on a boom like an Xi - you can always stop the case. Undock and move the whole boom out of the way. Or with modular carts - you can stop the case - undock the instruments and wheels the arms out of the way. And convert to a lap or open case.
But one inherent issue with bed mounted arms is a "cage effect" they create around the patient.

In this patent design - imagine that all the arms lock up. There is a critcal event like a major bleed. And now the team needs to open the patient - or perform a resuscitation.. (Rare but can happen). The cage around the patient makes the access way harder - you can't just roll the arms out of the way like a boom or carts.
The answer can be - that the arms never fault.
Or that if they do fault - the design never actualy locks the arms.
It could be that even in a mid priority alarm you can always tuck the arms away - you can always manually back drive them.
They may have several emergency solutions - but my natural concern (and having lived major alarms in say thoracic cases) is that this design adds some potential risks when you get into an emergency situation. Thoughts?
Replacing broken arms
Let me now be a little less dramatic. And let's say that one set of joints on one of the arms has an issue and one arm alone faults. (This happens with the boom robots as well by the way.)
With cart mounted systems - you simply undock that arm - wheel it out - and continue operating and bring in a new arm. Fire it up and continue - while the engineers can fix the other arm at their leisure. One big advantage of carts.
It also means that after that case the set of arms is still ready for the next case - you're out of the emergency and you have a full set of arms for the next elective case. With this Ottava design - if one arm is down - effectively the entire "table robot" is down until an engineer fixes it. The non- modularity (be it boom or table) has this downside.
With a boom robot - you can (if you're lucky enough to have a spare lying around) just wheel in a new boom.
But with the patient lying on the bed - even if you had a second Ottava as a back up - (which they should do for IDE hint hint). You can't just switch the robot around the patient. You have to move the patient from bed to bed. Un drape - re-drape - access lines, anaesthesia issues risk etc etc. It's a mess.
Now some of the Ottava engineers are going to say - but we know failures like that with Xi are extremely rare - data helps us predict it - so arm failures are going to be very rare for Ottava. So on our risk scoring sheets we have said that the impact is high - but the occurrence if very unlikely.
No!
You have to compare yourself to the failure rates with an S - as you are new and out of the box. I know arms have improved - joints and encoders are better - electronics are more reliable - software is better etc etc etc.
But I've lived this - custom arms with custom software - with custom instruments on production hardware (not highly controlled lab equipment) - that gets fluid and gunk in it from daily use (when they are tucked under the bed they are not draped and therefore less exposed to environmental "dirt" like fluid and dirt in prolonged use.) If you are true to your aim "beds" will be used for open cases of all kinds. Real world use.
So even if one arm goes down you have three others and could potentially continue the case - right ?: erm maybe not. What if the arm that goes down is the upper right arm and that is the one at the limits of reach for the access to the ports. In a cart based system you can replace in that exact posiition. Or move one of the less critical (retraction) carts to that position to continue. With Ottava (or the design I see) you cannot do that. It's going to get challenging fast in arm down situations. Prolonged anasesthsia is an adverse event let's not forget.
Now lets say that you do get to the end of the case - and you finished it with just three arms - then what about the next case. Are you going to fire up the system with just 3 arms and a broken arm? No. no one does that - trying a case with one broken arm tucked away.
In cart based system you have a fresh arm in - and you can start the next case. With Ottava you need a whole back up bed system. And in the early days that means lots of big spare beds on site. (The floor space argument might not feel so good now.)
Three arm positions on one side of the patient
I have been in hundreds of robotic cases in my time. With all types of surgical procedures. Often the carts or arm positioning is fairly even around the patient. But quite often all the carts or all the arms need to come from just one side of the patient. And often that is 3 arms. A scope and two instruments. I see this in renal surgery set ups, thoracic set ups and ventra hernia set up. It all depends on the user and the patient and the size of the pathology being treated. But a lot of the case (more than 15%) have required at one point or another three arms on one side of the table.
With carts - it's easy - you put three on one side and you are good.
With booms like an Xi you position the X boom and approach from one side with three arms.
I don't know how you do this with this picture?

Unless there is some major mechanism we can't see that allows all arms to come from just one side of the bed (there are early patents designs by VERB and Auris on this) then I don't know how you get three arms onto one side.
The engineering head on me says - well the arms are long enough that any arm can do a reach over to the other side - so it's no issue.
But I'm looking at joint lengths here and my experience is screaming - in a huge patient - with a large extended abdomen these arms will struggle to do reach overs and still have range of motion - joint angles for kinematics - pose positioning that will not alarm or clash.
Bottom line - it looks either very difficult - or impossible in some cases. And that's a concern.
The dance of moving systems
Workflow and OR efficiency are the buzz words for robotics. And on the face of it a "Bed robot" seems an easy proposition to move around and help with workflow. But 34 years of OR experience in every specialty surgeons do, in thousands of cases in over 50 different countries, hundreds of robotic cases. I'm struggling with how this bed design will help.
Let me talk logistics - and I'm coming at this from watching OR set up and tear down and scheduling workflows. The minute you LOCK the arms on the bed you create operational challenges. Do you drape the arms before the patient is brought into the OR? And again - draping and then refolding under knee height will not be acceptable in many countries. It is a risk of a contamination.
Alternative - bring the patient in - get them on the bed - then drape the patent - and then activate the arms out and drape them? Maybe - they will be simple to drape. But it is lost time as most systems are draped before the patient is draped to save time. This way wouldn't work.
You could drape the arms and not refold them - leave them out - but that leads to logistical issues of presenting the patient to the bed and to slide the patient onto the bed (transfer.) Because a major concern I have here is that in the USA the patient is induced and anaesthtised in the OR, on the actual opertating room table - that does not happen across the world. But if work flows are based on that idea - then a big surprise is coming in some places OUS. Where the patient is anaesthetised in an anti chamber or anaesthesia room while the prior case in the OR is finishing up. They get wheeled in already anaesthetised and then must be transferred to the operating bed.
There is an entire dance of OR turnaround that happens - and for me the "bed as the robot" in many parts of the world will break that flow. It will mean that you will need to have Ottava beds for every patient to keep a steady work flow - and that is just not feasible - and again then brings draping issues. (It's a long complex argument so sorry I'm not making it clear here.) But if usability has been done with OR teams just stateside - there is going to be a very rude awakening in some parts of the world.
Also some other things that concern me:
If this bed is being used for this case - where do I store my standard bed from the OR?
This bed might be needed in multiple ORs in a day - so you are not throwing out all your OR beds and replacing them with Ottava. Not happening - so that dance of beds in - beds out - move the beds around from OR to OR is just going to be something nursing staff will hate.
It's okay for some early cases in a fixed OR - but that is not the model that robots are going to.
I see this and I'm concerned about "one bed fits all" - how many versions will you eventually need to accommodate all patients and all procedure types?
What about the growing number of ORs that have fixed columns in the OR and the table top transfers across. Are those ORs now out of bounds?
Height adjustment of the arms - many cases you really need to come from a very high position and work down. In this image (and I can be wrong) there seems to be no way to pump up J1 to a very high starting place to allow a downward approach - critical. Look at both Hugo and Versius and see the height they need to get to. Da Vinci does it by design.
Locking the arms to the bed constrains that - especially as opposed to Zues or the DLR apporoach - both coming from patient height - both abandoned. The Ottava arms start at a lower position. Transoral will be a struggle if not impossible and many abdominal cases with high domed abdomens will (in my humble opinion) create all kinds of pose issues and leave those arms in horrible joint angles.
I fear for clash. If they use the common Z rail archiutecture (I hope not for their sake) then you limit the ability (unlike carts) to distance the J1s and therefore reduce clash angles. The more you restrain the J1 position the less choices you have in real world set ups. (maybe they have this figured out.)
Cost of good lokks very concerning here. These look very much like Versius style arms - which means bespoke joints and control systems. Not a lot off the shelf. I can't see the terminal joints - I can just see what I think is J1 to J6 - so the 7-8-9 will be interesting to see how much cost is in that - in the drive units.
But the bed integration and what is clearly a da Vinci style console look premium ( I expect nothing less from JNJ and a decade of delays).
But bespoke components and a bed - the COGS will be very high from what I see. The complexity looks very high. So that means building at scale will be vey hard. I think this will become a "aston martin" like build process in the early days - and no one is selling lots of this system for cash up front. You'll need back ups "ago-go" and servicing will be pretty bespoke and added difficulties of the bed.
All in all I predict they would need a $2.5 million selling price to make this work early days - and with limited procedures and end effectors - good luck if that's the case.
If they are in a "place it mode" where the market is going - this is a big hole they are about to dig in the P&L.
Console closed

So it's a closed console here in this shot - and it may be a DV5 style console. High end and emersive. But my concern is that arm management is going to be tricky in certain cases - and most modular systems need an open console to help with on the fly arm management.
IF this system behaves more like an upside down Xi - then it will be okay. But if kinematic wise it behaves more like a Versius - this console will become its bain. Hard for me to comment further until I see the arms in action.
Others have tried and abandoned
Although marketing speil will say that this is a "first". It's not.



Computer motion have tried it - Intuitive have tried it. Hell Medtronic bought DLR and tried it. CMR even tried it early on. And everyone one of them abandoned it as the real world understading of what a bed mounted robot means and it hits them. All ended up with a boom or carts. And for very good reasons.
Maybe the folks at Ethiocn have found that magic architecture - maybe technology is at a point where all the downsides of bed mounted can be overcome. Maybe they know something that other companies just don't about worklflow. Time is going to tell us and tell us soon. IDE is starting 2024 (end of) by their estimates.
I'm super excited about Ottava and I want it to succeed more than anything out of JNJ. But as my article says I'm concerned.
The potential benefits are massive - exciting - different. But maybe I have too much scar tissue in this so am a little sceptical. I'm hoping that JNJ launch Ottava and show me I am completely wrong. Because if they don't - in my opinion you can write off their endo-mechanical franchise. Let's all root for Ottava.
These are opinions of the author an simply for educational purposes. Written in a hurry in the middle of the night... sorry for mistakes.