T 891/07 – On The Sufficiency Of Humanoids

In the present decision the Board deals with the refusal of an application by the Examining Division (ED). The application was refused on the ground that the invention was not disclosed in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art (A 83). Additional comments were provided in the contested decision indicating that the ED was also of the view that the claims were not supported by the description (A 84).

Claim 1 on file before the Board read:

A system for gathering information from an environment, the system comprising :

- RFID tags to be provided on objects and transmitting information on the associated objects, and

- a humanoid robot provided with a computing device, the computing device comprising: visual sensing means for gathering visual information on objects in the environment, the environment being an area surrounding the computing device, wherein the computing device gathers said visual information and receives said information transmitted from said RFID tags of said objects located within said area, antenna means for wirelessly receiving said information from said RFID tags of said objects, means for combining the visual information as well as the wirelessly received RFID information for making a decision on objects identified by the visual information and the RFID information and for generating output signals based on said decision in order to drive a manipulator of the humanoid robot.

A 83 and A 84 (EPC 1973) - general remarks

[6.1] Both the requirement that the claims be supported by the description (A 84) and the requirement that the invention be sufficiently disclosed (A 83) are designed to reflect the principle that the terms of the claim should be commensurate with, or be justified by, the actual contribution to the art (T 409/91 [3.5]). It is the actual contribution to the art which defines the invention and for which protection may be sought.

[… T]he independent claims have been directed to the general concept of combining visual and RFID information. It is this concept which is currently considered to represent the contribution to the art.

[6.2] The Board supports the idea that an invention which opens up a whole new field is entitled to more generality in the claims than one which is concerned with advances in a known technology (Guidelines for Examination in the EPO, C-III, 6.2).

In view of the fact that multimodal integration was known at the priority date of the application (as can be seen from US-A1-2003/0149803, cited in the application), whether the present invention really opens up a whole new field may perhaps be questioned. Nevertheless, if the assessment of novelty and inventive step shows that the combination of visual and RFID information in a humanoid robotic context is new and inventive, then the Board considers that the appellant would be entitled to protection at this level of generality. This illustrates the interrelation that exists between A 84, A 83 and A 52(1) and highlights the risk that a broad claim will be more vulnerable to a novelty or inventive step objection.

A 83

[7.1] A 83 requires that the invention be disclosed in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art. For an application to be refused on this ground, the skilled person, equipped only with the description, claims and drawings of the application, and his general knowledge, would have to be unable to produce a working version of the claimed system. It cannot be expected that the skilled person will have to resort to inventive ingenuity in order to fill any gaps in the teaching of the application.

[7.2] Before it can be considered whether the skilled person would be able to carry out the invention, it is first necessary to establish what qualifications and level of knowledge the skilled person holds.

The Board accepts the position of the appellant that the skilled person in this particular field will be very highly qualified. In particular in the field of humanoid robotics, the skilled person will typically come from a background of computer science, but will require expertise in the fields of mechanics, electronics, biomechanics, computer programming and humanoid signal processing. Consequently, the “skilled person” in this field will comprise a team of experts, this team being technically very highly skilled and having a vast range of technical literature at its disposal.

[7.3] The invention, as it is set out in the independent claims, concerns the use of two signals, one emanating from the visual sensing means and one emanating from the RFID transmitter. These two signals are combined for making a decision on objects identified by the two information sources. The invention does not lie in the details of the manner in which the signals are combined, but instead in the mere fact that they are combined.

[7.4] The ED considered that A 83 was not satisfied since the application did not provide sufficient information to enable the skilled person to perform the combination of visual and RFID information without undue burden. No information was provided as to how the combination was performed, how the decision was taken, nor indeed as to the nature of the decision. No explanation of the specific processing required for multimodal integration was contained in the application, so the skilled person would not know how to perform the necessary data fusion. Even when referring to the specific example provided on page 5 of the application, no details were provided as to how the information integration should be performed.

Clearly, processing the visual and the RFID signals to make a decision will involve a certain degree of programming skill and signal processing knowledge: this task is certainly not trivial. It is noted however that data fusion in a multi-sensory environment is presented in the description as being the starting point of the invention […]. The skilled person must therefore be considered to be conversant with such data fusion techniques. Moreover, the description refers to US-A1-2003/0149803 which concerns multimodal integration. The skilled person may therefore also be considered to be conversant with multimodal integration techniques. Thus it must be assumed that the skilled person having the level of knowledge set out above will know how to process the data received from the two different sources and be in a position to draw a conclusion based on the combined data. The Board observes that, at a very basic level, this data could, for example, simply be a change in contrast boundaries sensed by the visual sensing means and information sent by the RFID tag establishing the identity of the object. As pointed out by the appellant, the decision which is taken need not even be correct. The Board is therefore convinced that the skilled person in this field will have sufficient knowledge to provide a “means for combining the visual information as well as the wirelessly received information for making a decision on objects identified by the visual information and the RFID information” and for generating a corresponding output signal.

[7.5] Moreover, the ED noted that the application contained no guidance on how the vision system formulated the hypotheses mentioned on page 5 of the description, nor did it instruct on a general framework of formulating hypotheses which could be applied in the context of heterogeneous data fusion. The ED therefore concluded that the

“essential components of the invention, in particular those concerning object recognition during the process of data fusion are missing from the application as filed” […]

and that the skilled person would have to resort to inventive activity in order to integrate the visual and the RFID data.

The Board notes that the invention as claimed does not require that any hypotheses be determined. Indeed, as contemplated above, depending on the decision to be taken and on the data acquired from the two sources, the combination of the visual and RFID information may in fact be a very straightforward operation. The Board is therefore of the view that details of the hypotheses-forming procedure are not categorically required in order to carry out the invention. Notwithstanding this finding, the appellant submitted that the skilled person in this field would nonetheless know how to form hypotheses for viewed objects. The Board tends to agree with this opinion and notes that the hypothesis need not be of a complex nature. For example, motion information from the visual sensor may be sufficient for the vision system to supply a hypothesis suggesting the possible identity of the object being viewed.

[7.6] The ED maintained that “taking a decision” in claim 1 must involve a process of object recognition or identification […] and held that, since no information was provided as to how to perform object recognition, the invention is insufficiently disclosed. The Board sees no basis for interpreting the “decision” in this manner. As argued by the appellant, the decision is unspecified and depends entirely on what information is received from the two sources.

[7.7] The ED further objected to the fact that the application did not contain sufficient details to enable the invention to be carried out over the whole breadth claimed […], basing this objection on the problem of correct data association in the specific scenario in which two objects are present (but not necessarily distinguishable as two objects) in the visual scene and two RFID signals are received. This scenario was covered by claim 1 but no explanation was provided as to how to correctly associate the two RFID signals to the respective visually-sensed objects.

A 83 requires that the invention be disclosed in such manner as to allow it to be carried out. A 83 does not require that details should be provided to enable each and every construction covered by the claims to be implemented; indeed, this would be an impossible undertaking when the claims are drafted in functional terms. In the view of the Board, even if hypothetical, non-implementable constructions may be conceived, this does not prevent the general principle being claimed in its full breadth if the skilled person is not put in a position where he systematically has to resort to inventive ingenuity in order to carry out the invention.

The example given on page 5 of the description is a very specific illustration of one scenario in which the invention could be employed. The invention does not hinge on whether two visually-sensed objects may be distinguished from each other, but that visual and RFID information may be used to augment each other and to permit decisions to be taken with little effort […]. It is in this breadth that the invention has been set out in the description and it is in this breadth that it is now claimed. Whether the skilled person will know how to analyse each and every scene presented to the computing device and to derive specific information from that scene is not essential for assessing whether A 83 is satisfied. What is important is whether the skilled person will know how to carry out the invention, which in the present case resides merely in the combination of visual and RFID information.

In the Board’s view, the examination of A 83 should not be reduced to a quest for a single construction which cannot be implemented on the basis of the information provided in the application. A 83 concerns whether the invention has been sufficiently disclosed rather than whether scenarios exist for which a full explanation has not been provided.

[7.8] In the present case, the Board therefore considers A 83 to be fulfilled.

A 84

[8.1] The ED held that “the description fails to support to a sufficient extent” the means for combining the visual information with the RFID information, since the application contained no clear teaching on how to successfully implement the invention when two objects were viewed and information was received from two RFID tags. The Board has understood this objection to be a lack of support by the description (A 84) over the whole breadth claimed.

Indeed, the RFID information and the visual information were initially not defined in the claims as pertaining to the same object. There was no teaching in the description which suggested that information other than object-centred information may be combined with the visual information, the implication being that the two types of information must relate to one and the same object. No other combination had been envisaged and no justification appeared to exist for generalising this concept to allow for the combination of visual information with non-related RFID information.

During the appeal proceedings claim 1 was amended and now sets out, using the wording of the original description, that the decision is taken on objects identified by the visual information and the RFID information. The Board considers that this serves to clarify that both information sources are used in the identification of the objects. That the objects may be identified by the visual and the RFID information means that enough information must be provided by each information source to enable the correct association of the data such that object identification is possible. Consequently, the claims have been restricted to the combination of information pertaining to the same object and so are now supported by the description in this respect.

[8.2] The Board acknowledges that the claims are still very broad, relating to the general concept of combining visual information about an object with RFID information about that object. However, the Board does not consider that this generality goes beyond what is warranted by that which the appellant presents as being the contribution to the art and which is described in the application as the mere combination of the visual and the RFID signals. As will be clear from the remarks concerning A 83 above, the Board considers that the invention may currently be seen to lie at this level of generality.

In view of the fact that novelty and inventive step have not yet been considered - this being the subject of the examination still to be performed by the ED - any limitation at this stage of the proceedings to more specific embodiments would be unduly restrictive.

[8.3] Consequently the Board is of the opinion that the claims are supported by the description.

The Board then remitted the case to the ED.

Should you wish to download the whole document, just click here.

To have a look at the file wrapper, click here.