How Smart is your Cleaning Robot?

by Dennis Crouch

iRobot Corp. v. USITC (Fed. Cir. 2019)

Back in 2017 iRobot (maker of the Roomba) complained to the USITC about infringing autonomous vacuums being imported into the US.  The patentee won on some grounds, but lost on others.

This appeal focuses on iRobot’s U.S. Patent 9,486,924 which allows for scheduling of the cleaning via mobile phone.  After conducting its investigation, the ITC found no infringement.  A divided appellate panel has affirmed.

The issue on appeal is claim-construction.  The claims include a requirement that a phone user transmit certain “instructions” to the cleaning robot.

1. A method of cleaning a room, the method comprising:

transmitting from a cleaning robot to a mobile phone a status of the cleaning robot; and

receiving at the cleaning robot from the mobile phone, in response to an operator command input at the mobile phone and at least in part indicative of a schedule, information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation in the room according to the schedule,

wherein executing the cleaning operation in the room according to the schedule comprises:

leaving a stationary charging device at which
the cleaning robot is docked according to the
schedule, and

navigating about a floor surface of a room.

The question in the case is whether those “instructions” can simply be schedule information, or must they be machine executable code.

iRobot argues that its proposed construction could include, for example, a simple instruction to begin the cleaning cycle.

On appeal, the Federal Circuit rejected that approach and instead held that “instructions” must be computer executable code. The patent here does not define “instructions” one way or the other and so perhaps this is an OK conclusion albeit somewhat narrow.  The court notes that the claims at issue here are considerably narrower than those found in a parent application — those original claims did not include the “instruction” limitation.  One problem not addressed here

The majority opinion was penned by Judge Dyk and joined by Judge Mayer.  Judge Bryson wrote in dissent:

The problem resides principally in a single claim term, the word “instructions.” Unfortunately, that term appears for the first time in the claims (except for two usages in an entirely different context in the specification), and the meaning of the term is therefore not immediately apparent. . . . I would read the term as having a non-technical meaning such as “directions,” “commands,” or “orders.”

Judge Bryson’s approach here fits better with my reading of the patent document itself since there is no enablement or description in the patent itself of how the phone would be sending computer code to the iRobot.  Still, I don’t have much sympathy to a patentee in this case who added new and not-defined limitations into a continuation application.

= = = = 

At its core, the dispute between the majority and dissent is about the intelligence of the “cleaning robot.”  If the device is an AI-enabled robot then Bryson’s opinion makes lots of sense because the device will be able to understand lots of commands.  On the other hand, a lower-level electronic device may need more technical instructions in order to guide its progress.

= = = =

Impact of the provisional: In this case, the patent at issue is a continuation from a prior utility application and also claims priority to a prior provisional application.  Neither of the non-provisional applications talk through the process of sending machine-code from a phone to the Roomba.  However, the provisional does go into depth on an approach to transfer new code and algorithms to the vacuum from a separate device or drive.  Those elements of the invention were seemingly stripped-out of the non-provisional before it was filed.   The opinion does not mention the provisional application, but the situation begs to be addressed.  At first glance, the provisional provides a strong indication that the inventors contemplated the idea of transferring machine code instructions from a remote device to the vacuum — supporting the majority opinion.  On the other hand, the fact that those disclosures were removed from the non-provisional application indicates that the ‘924 patent was moving in a different direction.

66 thoughts on “How Smart is your Cleaning Robot?

  1. 12

    I think their interpretation of “instructions” is probably too narrow, but worse it’s sloppy. Equating “instructions” to “program or machine executable code” is awful as “program” can refer to either machine executable code or a schedule.

    Anyways…

    The specification clearly discusses transmitting both executable code and non-executable scheduling information. So the patentee actually had solid support for the invention they needed! Hah!

    1. 12.1

      >at least in part indicative of a schedule, information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation in the room according to the schedule,

      Ben, I think the language “configured to cause a processor …. ” indicates they are code not an upper level command.

      Normally, if it was a claim to a command the claim would read something where the receiver does something in response to receiving the command.

    2. 12.2

      >>information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation in the room according to the schedule,

      Usually this type of language indicates computer code. Does an instruction like “wake-up” “configure[] to cause a processor of the cleaning robot …”. Not really.

      Again–if it was something like “wake-up”, then the claim should have been written differently.

      You see the claim should have been written like:

      “in response to receiving an instruction from a mobile phone, blah blah …”

      You get it? The “instruction” as written is tied to the processor.

      1. 12.2.1

        “Does an instruction like “wake-up” “configure[] to cause a processor of the cleaning robot …”. Not really.”

        I just don’t see that. A “go clean” command absolutely causes the processor to execute the processes associated with “cleaning operations.”

        Of course the claim could have been written better. That’s the fun of this case. They had great support for A, B, and (A or B), but played around with the language and it cost them.

        1. 12.2.1.1

          Ben, you did not address my point.

          The command “wake-up” is not configured for the processor.

          1. 12.2.1.1.1

            I’m sincerely trying to not be dense, but I don’t get your point. One last try?

            If you have a processor arranged to receive a rising edge TTL signal and responsive to receiving that TTL signal executes a set of operations, is the TTL signal not an “instruction configured to cause a processor” to execute those operations? If not, why not?

            If the signal isn’t configured correctly, the processor won’t react. And yet, such a signal is definitely not executable code.

            1. 12.2.1.1.1.1

              Ben,

              I’d say yes on your TTL signal.

              Let’s look at this claim again.

              “at least in part indicative of a schedule, information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation in the room according to the schedule”

              The terms “configured to cause the processor … to execute a cleaning operation in the room according to the schedule.”

              I think the problem with saying that a command like “wake-up” is “configured to cause the processor” is that the command “wake-up” isn’t really related to the processor but to the software that would receive the command. The software running on the robot is configured to cause the processor to do things, but the command “wake-up” really isn’t.

              This is a bit funny in that my view is one of seeing the software + processor as a new machine. The command “wake-up” may cause the new machine to do something, but the command is independent of the processor. The processor could be switched to a different processor that operates completely differently and the command “wake-up” would not know the difference, so how can it be configured to cause the processor to do anything?

              In fact the entire machine could be switched out and the command “wake-up” could still work.

              One skilled in the art that looks at the term instructions and configured to cause a processor to do something thinks of instructions that operate the processor.

    3. 12.3

      And actually the more I look at the claim the more I think my position is right.

      A command like “wake-up” is not configured to cause a processor to do something. It is configured to cause the robot to do something. There is no tie between the command “wake-up” and the processor on the robot.

      1. 12.3.1

        Your comment here is in line with my original post at 7.

        However, on second look I modified that position because the claim language has a “flexible” ‘including’ use, which may “flex” just how the “configured to” phrase operates.

        The phrase “information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation

        In one version, the information has to be of the type that can configure a processor to do something.

        In another version, the information includes a something else, with that something else being the item that is of the type that can configure a processor to do something.

        Ah, use of commas may have been helpful here.

        1. 12.3.1.1

          >In another version, the information includes a something else, with that something else being the item that is of the type that can configure a processor to do something.

          What other version are you talking about?

        2. 12.3.1.2

          “the information has to be of the type that can configure a processor to do something”

          “the item that is of the type that can configure a processor to do something”

          What limitation requires “configuring” the processor? I only see “causing” the processor to do something.

          1. 12.3.1.2.1

            the phrase “configured to” indicates that at least one of either the first noun in the series (information) or the second noun in the series (instructions) are to be of the type that can configure a processor to do something.

            To answer your question, the phrase I presented is NOT a requiring of configuring the processer (while “causing” in your terms is easily recognized as doing that configuring).

            If anything, the “require” may be deemed implicit in the “according to the schedule” part of the claim (but I am not hung up on even that).

            1. 12.3.1.2.1.1

              “at least in part indicative of a schedule, information including instructions configured to cause a processor of the cleaning robot to execute a cleaning operation in the room according to the schedule,”

              ??? What? Something has to be “configured to cause a processor”.

              Their argument is that “wake-up” fits “configured to cause a processor.” It doesn’t.

              1. 12.3.1.2.1.1.1

                Night Writer,

                In the version in which it is only the second noun in the series to which THAT noun is to be of the type that can configure a processor to do something, the first noun then can be something other that the thing that is to be of the type that can configure a processor to do something.

                1. I get what you are saying anon.

                  But then the information would have to include the instructions and the command.

                  I don’t think this helps their case.

                2. You are probably correct that this may not help their case, but I was more interested in the claim construction of serial nouns and a descriptor that in combination may provide multiple (non-identical) meanings.

                  Is this indefinite or merely broad? Either very well may not help.

  2. 11

    “I would read the term as having a non-technical meaning such as “directions,” “commands,” or “orders.””

    That’s cool bruh but how would an imaginary man of ordinary skill in the art read that term?

    1. 11.2

      Since the specification does not even use “instructions” in any kind of context I would read “information including instructions configured to cause a processor of the cleaning robot” as an element claimed only by its function, where “information including instructions” is a nonce word equivalent to “means” or “stuff.” I doubt the drafter viewed the format or nature of the “information including instructions” to be germane to patentability and chose a deliberately broad term to encompass anything except information that doesn’t include instructions.

  3. 10

    Folks, I see that a significant event passed here a few weeks ago with no mention. The 15th anniversary of this fine blog arrived on 9 Apr 2019. I think that we should all take a moment to congratulate Profs. Crouch & Rantanen on their many years of excellent work here.

    Happy belated anniversary to the finest patent law blog on the web!

  4. 9

    Sending a command to something to perform an action is old as dirt. The content of the command or “instruction” is a meaningless distinction, just like the subject matter of data being processed is meaningless.

      1. 9.1.1

        It is more than merely context.

        Ordinary’s quip here is akin to wanting the judicial doctrine of printed matter WITHOUT taking the exceptions to that doctrine.

        It is also very much against the XKCD panel on tasks ( link to xkcd.com ), as well as aligned with the vapid “House” quip that once a programmable computer was invented, then ALL future improvements by way of the programming are somehow “already in there.”

        It is mindlessness of the most vapid sort.

      2. 9.1.2

        I agree that context can be important. However, I do not understand (from the claim) what makes this interaction between a phone and a device any different than any other interaction between a phone and a device where the phone sends a command to the device.

        1. 9.1.2.1

          “I do not understand (from the claim) what makes this interaction between a phone and a device any different than any other interaction between a phone and a device where the phone sends a command to the device”

          The answer is: nothing.

          The context is: we had cleaning robots, we had charging docks, we had mobile phones which send commands to robots (i.e. US6845297) and we had schedules.

          1. 9.1.2.1.1

            The context is: we had cleaning robots, we had charging docks, we had mobile phones which send commands to robots (i.e. US6845297) and we had schedules.

            But these schedules included night time events with odd numbered hours.

    1. 9.2

      What you are doing is trying to abstract away the details which is never OK in real patent law.

      When is something sent? Why is it sent? What is sent? What is the response to sending it? Etc.

      1. 9.2.1

        What you are doing is trying to abstract away the details which is never OK in real patent law.

        Maybe somebody should tell that to the Supreme Court.

        1. 9.2.1.1

          Hmmm,

          Was that “prescriptive” or “descriptive?”

          Does “real patent law” know the difference?

  5. 8

    Isn’t he relationship, or lack of it, between ITC proceedings and IPRs separately trying patent validity and claim scope issues of the same patents something Congress really ought to be looking into?

    1. 8.1

      I thought that the reason for the existence of the IPRs was to provide a “lack of relationship” (different alternative path) from other avenues…

      1. 8.1.1

        The main reason for both reexaminations and IPRs is to greatly reduce litigation trial and appeal costs, and the shake-down settlement costs caused by that, of patent suits on patent claims that are not valid over even prior patents and publications. ITC patent proceedings can also have very high pre-trial litigation costs, and they typically accumulate even more rapidly.

        1. 8.1.1.1

          so…

          Your reply here answers your first question (as I indicated).

          Glad that you could help yourself.

  6. 7

    Comments so far have had the wrong focus.

    The focus on “instructions” per se misses the actual claim language of “instructions configured to cause…”

    There is more dust here than substance (talk about ‘nothing burgers’).

  7. 6

    In this case, technical isn’t even really the issue. That is a word game being played. The real issue in this case is whether the “instruction” was meant to be a command (e.g., start this, do that, etc.) or software to perform the command. I think the majority is right that the claim language and specification indicate that what was meant by “instruction” in the claims is software to perform the action.

    Although, I didn’t look at the prosecution history. I note that both the majority opinion and the dissent have large problems in understanding the technology.

    We need judges with the proper education.

  8. 5

    BRYSON, Circuit Judge, dissenting.
    >I would read the term as having a non-technical meaning such as “directions,” >“commands,” or “orders.” The difference leads to a different construction of >independent claims 1 and 12 of the ’924 patent and a difference in outcome for >this appeal. I therefore respectfully dissent.

    So, let’s see here, a machine is receiving something from another machine and Bryson has decided that this is “non-technical.” Unbelievable.

    1. 5.1

      >>Department of Justice from 1978 to 1994
      >>A.B. from Harvard College in 1969

      This explains a lot. Why not just say, “I burn down any patent I can get my grubby hands on.”

      Bryson, you are a disgrace. Resign in shame. Two machines communicating is “non-technical.” It is tiresome to even think about such a statement it is wrong on so many levels.

      1. 5.1.1

        Two machines communicating is “non-technical.”

        I think the issue is whether the alleged improvement is “technical.”

        Consider the following:

        Prior art teaches an answering machine for recording and playing back messages.
        Prior art teaches a tape recorder for recording and playing back messages.
        I claim: “A method of duplicating a recording, comprising receiving recorded signals from an answering machine, wherein said receiving is by a tape recorder, and wherein said signals comprise real estate information.”

        Is that “technical improvement” to the tape recorder? It’s useful information and storage is a practical utility. But is that a “technical improvement”?

        It’s 2019, friend. Most of us recognize that two computers “communicating” with each other is older than the hills. This is a junk claim, written by bott0m feeders and asserted by people who are worse than bottom feeders.

        1. 5.1.1.1

          MM, I think it is a technical improvement to answer your question. It is using machines and so must be technical.

          I don’t think an enquiry into whether it is technical or not is useful. In this case, Bryson is actually just playing word games and I think he reached his result because he doesn’t really understand technology.

          And, in fact, in this case, technical isn’t even really the issue. That is a word game being played. The real issue in this case is whether the instruction was meant to be a command (e.g., start this, do that, etc.) or software to perform the command. I think the majority is right that the claim language and specification indicate that what was meant by “instruction” in the claims is software to perform the action.

          What I think is that 102/103/112 are all we need to police the patent system.

      1. 5.2.1

        “Why not just say, “I burn down any patent I can get my grubby hands on.” Bryson, you are a disgrace.”

        “Oh never mind…”

        You can do better than that! Maybe something along the lines of “Contrary to my immediate assumption, Bryson’s interpretation was favorable to the patentee. I’ll try to finish reading posts before calling a judge a disgrace in the future”?

        1. 5.2.1.1

          Except I agree with the majority and do not agree with BRYSON. I suspect that his dissent is based on his inability to think about technology.

        1. 5.2.2.1

          Except I agree with the majority and do not agree with BRYSON. I suspect that his dissent is based on his inability to think about technology.

  9. 4

    Also relevant prior art that should have tanked this claim: a zillion descriptions in the prior art of mothers on the phone with their sons where the mother tells the son to clean their rooms before the mother gets home, and where the son says he’s hungry and the mother tells him there’s an apple in the fridge.

    Only the USPTO pretends that everybody was born yesterday just because a shiny computerized robot is involved.

    News flash: computerized devices have been receiving instructions remotely since before most Examiners at the PTO were born. The US Patent System is a f cking joke and the only beneficiaries are the Silly Con Valley Cl 0wns. Pull the plug on this nonsense already.

  10. 3

    the Federal Circuit … held that “instructions” must be computer executable code

    Shocking! The software community must be extremely upset by this!!! Everybody knows that software isn’t instructions. Why, we’ve been told that a zillion times by … oh, never mind.

    1. 2.2

      The first Roomba (released in 2002) is prior art for a cleaning robot, the Ruffner reference (US6611738) teaches a task robot automatically using a charging doc. What’s left? Sending scheduling orders from a phone and the agent acting according to those instructions. Non-obvious? Please.

      Prosecution history is pretty embarassing too. Non-final with post-dated art. Applicant points out the date issue. Examiner throws up their hands and allows.

      1. 2.2.2

        Sending scheduling orders from a phone and the agent acting according to those instructions. Non-obvious? Please.

        Nobody expected that vacuuming robots could process instructions. They are notoriously stubborn! Even worse than hair drying robots.

    2. 2.3

      After a brief look at the history, I think the examiner probably dropped the ball in this case. But that doesn’t relieve the system of some fraction of the blame. We could have a system with higher fees and reduced time constraints for examination. That system would be less likely to produce this outcome. That system would be more conducive to greater examiner accountability. And we will never get that system as long as the largest group of PTO stakeholders act like they have no influence over how the PTO operates.

      1. 2.3.1

        That sounds in the classic “Wah, don’t blame me for not doing my job”

        It also sounds like you trying to confuse the metrics of your job with what your job actually is.

        No amount of any outsider’s “influence” will help you understand what you need to understand.

  11. 1

    Are “C3PO, shut down all the trash compactors on the detention level!” and “Open the Pod Bay doors, Hal” 103 prior art if combined with an electric tooth brush ad?

    1. 1.1

      “C3PO, shut down all the trash compactors on the detention level!” and “Open the Pod Bay doors, Hal” 103 prior art

      Descriptions of computerized apparati receiving instructions are most definitely relevant to a claim describing … a computerized apparatus receiving instructions.

      Why on earth would that not be the case? At the very least we have evidence of motivation to combine those functionalities (as if any were needed).

      1. 1.2.1

        Thanks for the link, Bguy.

        What the article does not do (unfortunately) is actually distinguish the fact that science fiction (or any fiction) may contain elements that are NOT fiction.

        It is the non-fiction elements that serve a purpose that actual “science fiction” may not.

        And yes, enablement is an important differentiator (albeit, your link really does not go into that aspect). One important aspect to include here is that certain prior art items DO contain a presumption of enablement, while other prior art items contain no such presumption.

        1. 1.2.2.1

          E.g, Fed. Cir. Amgen Inc. v. Hoechst Marion Roussel Inc.:
          “Obviousness – The Sugimoto Patent
          The district court held in Amgen I that the Sugimoto patent did not render claim 1 of the ‘422 patent obvious because the Sugimoto patent was not enabled. 126 F.Supp.2d at 114 & n.29. In Amgen II we vacated this finding and remanded because nonenablement does not preclude a finding of obviousness. 314 F.3d at 1357.”

          1. 1.2.2.1.1

            That appears to be a misapplication of the details of the case, Paul.

            the case under at 1355 indicates “We indicated that the Sugimoto patent should have been presumed enabled and that Amgen should have had the burden of proving otherwise.

            The presumption of enablement is tied to the granted patent – and the use of the granted patent for obviousness did not turn on the issue as in that case of nonenablement. It is an error to take the case and try to expand it to have ALL of obviousness/non-obviousness and enablement for ANY reference (specifically NPL and even more specifically science fiction NPL).

            At least that’s my take from: link to finnegan.com

          2. 1.2.2.1.2

            Here’s a better view:

            link to casetext.com

            The “presumption” of enablement indeed appears to have been expanded beyond the (reasonable) view that a reference has been vetted by the Office in a grant of patent to most any publication, HOWEVER, that presumption is removed with but a challenge that offers any solid reasoning. Once so challenged, the onus switches back to the Office to show proper enablement.

Comments are closed.