Method for assembling loose parts.
The method and system for assembling bulk parts using force monitoring during the assembly process address inefficiencies by ensuring secure and efficient fitting without anti-collision interference, enhancing the assembly process.
Patent Information
- Application Number
- FR2022012432
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-11-28
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-11-28
AI Technical Summary
Existing industrial assembly processes face inefficiencies in assembling unordered bulk parts due to the need for separate robots and anti-collision mechanisms, which complicate the assembly process and increase time, especially when parts are tangled or overlapping.
A method and system using a robot with actuated joints and a gripping member to assemble parts by monitoring force variations during the assembly process, ensuring the force experienced conforms to an expected variation representative of a successful assembly, allowing simultaneous handling and assembly without anti-collision interference.
Enables efficient and secure assembly of bulk parts by ensuring correct fitting without sacrificing safety, reducing time and complexity by integrating force monitoring with anti-collision mechanisms.
Smart Images

Figure 00000022_0000 
Figure 00000023_0000 
Figure 00000024_0000
Abstract
Description
Title of the invention: Method for assembling loose parts. [0001 ] GENERAL TECHNICAL FIELD
[0002] The present invention relates to the field of robotization of industrial processes and in particular of production lines. The present invention relates more specifically to an assembly method with a picking component, that is to say, gripping parts arranged in bulk in a container while assembling them directly.
[0003] STATE OF THE ART
[0004] Parts intended for use in manufacturing lines are generally supplied "in bulk", that is to say arranged in a disorganized manner in a container, as opposed to "unit" packaging in which each part is stored individually.
[0005] Bin picking is the highly repetitive and non-value-added task of picking them up one by one to place them elsewhere and then use them.
[0006] There have recently been attempts to automate this task, in particular by using robots with vision capabilities and gripping means. More precisely, a robot arm picks up the parts one by one from the container. The difficulty is that the container can contain several hundred or thousands of parts that may be tangled and overlapping, which is why it is necessary to use an advanced algorithm for locating parts arranged in bulk, as described in application FR2204180.
[0007] This picking technique is entirely satisfactory but currently only allows a part to be picked up and placed somewhere, generally a buffer zone. If one wishes to assemble this part with another, and in particular to fit one into the other, it is necessary to use another dedicated robot, generally a press applying force to the parts, resulting in a loss of time.
[0008] Indeed, the robot arms used for bin picking implement an anti-collision mechanism which makes it possible to stop the movement automatically if an actuator detects a resistance force greater than a threshold (characteristic of a contact), so that it is impossible to envisage applying a force with the arm, hence the use of dedicated presses. However, this anti-collision mechanism is essential for safety reasons, because it is necessary to be able to interrupt the movement of the robot in the event of an unforeseen obstacle in the path of the arm.
[0009] At best it has been proposed that the robot arm changes jaws (precision jaws, pressing jaws) after having placed the part on the buffer zone, or even directly presents a press tool not subject to T anti-collision, to carry out the force, which complicates
[0010]
[0011]
[0012]
[0013]
[0014]
[0015] the task without really speeding it up. The present invention improves the situation. PRESENTATION OF THE INVENTION The present invention therefore relates, according to a first aspect, to a method of assembling at least a first part with a second part, the first part being arranged in bulk in a first container, the method being characterized in that it comprises the implementation by data processing means of steps of: a. Determination of a first trajectory of a first robot equipped with actuated joints and a gripping member, making it possible to grasp said first part in the first container with the gripping member of the first robot and then to move it until it is assembled with the second part, said second part being held by a gripping member of a second robot different from the first robot in a predefined waiting position; b. Control of the first robot so as to implement said trajectory, by estimating a force undergone by said actuated joints of the first robot as a function of the progress of the trajectory; c. Verification that during the trajectory: - Until contact between the first part and the second part (2b), the force undergone is less than a safety threshold; - then, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part with the second part. According to advantageous and non-limiting characteristics: The second part is also arranged in bulk in a second container, the method comprising prior steps of: a. Determination of a second trajectory of a third robot equipped with actuated joints and a gripping member making it possible to grasp said second part in the second container with the gripping member of the third robot and to move it to said waiting position; b. Commanding the third robot so as to implement said second trajectory. The third robot is the second robot. The third robot is not the second robot, preferably is the first robot, and step (b) comprises transferring said second part from the gripping member of the third robot to the gripping member of the second robot.
[0016] Step (c) comprises the prior localization of said first part in the first container based on at least one three-dimensional point cloud of the parts arranged loosely in the first container, acquired by a stereoscopic camera.
[0017] The first trajectory of the first robot successively comprises: An initial phase, in which the first robot moves from a starting position to a gripping position in which the gripping member of the first robot can grasp the first part; An approach phase, in which the first robot moves, holding the first part, from the gripping position to the contact position of the first part with the second part; A phase of fitting the first piece with the second piece.
[0018] The fitting phase comprises a first sliding sub-part and a second assembly sub-part, step (e) comprising verifying that: in the sliding sub-phase, an increase in the force undergone is beyond a minimum threshold but without reaching an intermediate threshold; In the assembly sub-phase, a maximum threshold greater than the intermediate threshold is reached within a predetermined interval of advancement position of the first robot.
[0019] The fitting phase comprises a first sub-phase of sliding to a depositing position in which the gripping member of the first robot can release the first part, and a second sub-phase of assembly by applying a thrust to the first part with a support surface of the first robot.
[0020] Step (e) comprises: If, before the contact of the first part with the second part, the force undergone reaches the said safety threshold, each robot is stopped; If, after contact of the first part with the second part, the variation in the force undergone is not in accordance with the expected variation representative of the assembly of the first part with the second part, the result of the assembly is rejected.
[0021] The method includes additional steps of: Determination of a third trajectory of the first robot allowing to deposit the assembly result to another container according to the result of the verification of step (e); Command of the first robot so as to implement said third tra jectory.
[0022] According to a second aspect, the invention relates to a system for assembling at least a first part with a second part, the first part being arranged in bulk in a first container, comprising a first robot provided with actuated joints and a gripping member, a second robot different from the first robot provided with a gripping member, and data processing means; characterized in that the data processing means are configured to: - determining a first trajectory of the first robot making it possible to grasp said first part in the first container with the gripping member of the first robot and then to move it until it is assembled with the second part, when said second part is held by the gripping member of the second robot in a predefined waiting position; - Control the first robot so as to implement said trajectory, by estimating a force undergone by said actuated joints of the first robot as a function of the progress of the trajectory; - Check that during the trajectory: • Until the first part comes into contact with the second part, the force experienced is less than a safety threshold; • then, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part with the second part.
[0023] According to advantageous and non-limiting characteristics, the system comprises a third robot provided with actuated joints and a gripping member, the second part being arranged loose in a second container (3b); the data processing means are configured to: - determining a second trajectory of the third robot making it possible to grasp said second part in the second container with the gripping member of the third robot and to move it to said waiting position; - Command the third robot so as to implement said second trajectory.
[0024] According to a second and a third aspect, there is provided a computer program product comprising code instructions for executing a method according to the first of assembling at least a first part with a second part, and a storage means readable by computer equipment on which is recorded a computer program product comprising code instructions for executing a method according to the first aspect of assembling at least a first part with a second part. PRESENTATION OF FIGURES
[0025] Other features and advantages of the present invention will become apparent from the reading the following description of a preferred embodiment. This description will be given with reference to the attached drawings in which:
[0026] [Fig.l] [Fig.l] is a diagram of a system for implementing the method according to the invention;
[0027] [Fig.2] [Fig.2] is a flowchart representing the steps of an embodiment of the method according to the invention;
[0028] [Fig.3a] [Fig.3a] illustrates a first state of the system in an embodiment of the method according to the invention;
[0029] [Fig.3b] [Fig.3b] illustrates a second state of the system in an embodiment of the method according to the invention;
[0030] [Fig.3c] [Fig.3c] illustrates a third state of the system in an embodiment of the method according to the invention;
[0031] [Fig.4a] [Fig.4a] represents a first example of a force / displacement curve used in an embodiment of the method according to the invention;
[0032] [Fig.4b] [Fig.4b] represents a second example of a force / displacement curve used in an embodiment of the method according to the invention. DETAILED DESCRIPTION
[0033] Architecture
[0034] With reference to [Fig.l], the present invention relates to a method for assembling at least one first part 2a with a second part 2b, the first part 2a being arranged loose in a first container 3a (which means that there are a plurality of copies of the first part 2a in said first container 3a), and advantageously the second part 2b is also arranged loose in a second container 3b (which means that there are a plurality of copies of the second part 2b in said second container 3b), even if it is possible that the second part 3b is already packaged in an orderly manner. It is recalled that by arranged “loose” is meant arranged in an unordered and generally random manner: when the containers are opened, the position of the parts inside and their interlocking is unpredictable.
[0035] The first and second containers (or containers) 3a, 3b are typically crates, opened so as to access the parts inside and grasp them individually. Note that it is possible for the first and second containers 3a, 3b to be combined, i.e. the first and second parts 2a, 2b are arranged loosely in the same container. In the example of [Fig.l] there are 4 containers: the first container 2a, the second container 3b, a third container 3c for receiving the assembled parts, and a fourth container 3d for receiving the rejected parts (see below).
[0036] It is understood that the method thus comprises both a stripping component (of at least the first part 2a), and an assembly component, both being carried out simultaneously, without going through a phase of repositioning the stripped parts.
[0037] By assembly, we mean the union of the two parts 2a, 2b so as to form a single part denoted 2a+2b. It is assumed in the remainder of the description that the parts 2a, 2b are assemblable, that is to say that they have complementary shapes. The parts 2a, 2b could also be identical, subject to their possible assembly.
[0038] We generally have a “male” part, i.e. with a protuberance (in the examples shown, it is the first part 2a, an axis) and a “female” part, i.e. a bore (in the examples shown, it is the second part 2b, a barrel), and the assembly is an engagement of the male part in the female part (more precisely of the protuberance in the bore), or “fitting”. The assembly can be force-fitted, and held by friction, or involve elastic deformation, clipping, etc.
[0039] The system for implementing the present method comprises data processing means 4 and at least two robots 1a, 1b (and potentially three). In the example of [Fig.l], there is only the first robot 1a and the second robot 1b. The reference 1c will be used for the possible third robot, but it is understood that in the examples shown the first robot 1a plays the role of the third robot. Alternatively, the second robot 1b can play the role of the third robot 1c; all possible configurations will be explained later.
[0040] The data processing means 4 are typically a processor, and are in particular intended to control the robots 1a, 1b, 1c. These are generally the data processing means of a computer, such as a laptop, connected to the robots 1a, 1b, 1c, which may further comprise. The system also typically comprises data storage means 5 (a memory, again potentially that of the PC), an interface 6 (for example a touch screen), and a structure 7 for supporting the robots 1a, 1b, 1c, in particular having a tray for placing the containers 3a, 3b, 3c, 3d and on which the interface 7 may be mounted.
[0041] The system further advantageously comprises a stereoscopic camera 10, also connected to the processing means 4. The camera 10 is arranged so as to see the interior of the containers and therefore the parts 2a, 2b, i.e. either fixedly or mounted on the first robot 1a (case of [Fig.l]), and in all cases capable of acquiring a three-dimensional point cloud, and generally a two-dimensional image. In the case of a third robot 1a distinct from the first robot 1a (see below), it may have its own camera 10.
[0042] Robots
[0043] The first robot 1a is a robot for handling at least the first part 2a. It is provided with actuated joints 12a and a gripping member 11a, and advantageously with a support surface 13a (see below).
[0044] Such a robot is generally called a "robot arm", because the actuated joints 12a, or "joints", allow it to change position and thus move a part 2a. In the example of [Fig. 1], the first robot is of the Staübli TX2-90L type with 6 actuated joints 12a, but any other architecture will be possible, including Cartesian robots, because the notion of actuated joints will be interpreted broadly.
[0045] The gripping member 11a is typically a 2-jaw clamp, operable independently of the joints, so as to “grasp” or “release” a part 2a, and generally located at the end of the arm so that all the joints 12a contribute to moving this gripping member 11a. The camera 10 is typically also at the end of the arm, so as to move in a manner integral with the gripping member 11a.
[0046] For convenience, in the remainder of the description, the value of the vector of the positions of each joint, denoted [jl, j2, j3, j4, j5, j6] in the case of the 6-axis robot, will be designated as the "position" of the first robot la. It is assumed that a dynamic model of the robot is available that makes it possible to match a spatial position [X, Y, Z] (in a given frame of reference - in particular orthonormal) of the gripping member 1 la with a position [jl, j2, j3, j4, j5, j6] of the robot la.
[0047] It is assumed that the first robot 1a implements a conventional anti-collision mechanism, i.e. that it is capable of estimating a force undergone by said actuated joints during a trajectory, in particular to detect an abnormal force representative of an impact against an obstacle (force greater than a safety threshold): the first robot then stops instantly.
[0048] The second robot 1b does not necessarily have actuated joints 12b, but has its gripping member 11b for gripping / releasing the second part 2b. In the examples shown, this is a 3-jaw gripper. It is assumed that a waiting position is predefined for the second robot 2b, which is fixed in the absence of actuated joints 11b.
[0049] Alternatively, the second robot 2b may be a complete robot arm (i.e. with actuated joints 12b), for example the same as the first robot 1a (not shown). It may then have its own stereoscopic camera.
[0050] Note that the first and second robots 1a, 1b are two distinct robots (in that they can be manipulated independently so that the first robot 1a can hold and move the first part 2a relative to the second robot 1b, i.e. they have different actuated joints and gripping members 11a, 11b), but they can "be part" of the same large robot.
[0051] As explained, the present method involves in particular at least the unpicking of the first part 2a by the first robot 2a. In the case where it also involves the unpicking of the second part 2b, this unpicking is done by a third robot 1e.
[0052] According to a first embodiment, there is a third robot 1a distinct from the other two robots 1a, 1b which is typically the same model as the first robot 1a (a robot arm with actuated joints and gripping member). The third robot 1a unpacks the second part 2b and gives it to the second robot 1b (which can be a robot without actuated joints), and the first robot 1b unpacks the first part 2a and implements the assembly, the two can unpack simultaneously.
[0053] According to a second embodiment, the third robot 1c is the second robot 1b, i.e. the second robot 1b also plays the role of the third robot and unpacks the second part 2b. It is then typically the same model as the first robot 1a (a robot arm with actuated joints and gripping member). The second robot 1b unpacks the second part 2b and places itself in the waiting position, then the first robot 1b unpacks the first part 2a and implements the assembly.
[0054] According to a third embodiment corresponding to what is represented by [Fig.l], the third robot 1c is the first robot 1a, i.e. the first robot 1a also plays the role of the third robot and unpacks the second part 2b. In other words, the same arm unpacks the two parts 2a, 2b. The first robot 1a unpacks the second part 2b and gives it to the second robot 1b (which can be a robot without actuated articulation), then the first robot 1b unpacks the first part 2a and implements the assembly. The two unpackings are then necessarily sequential.
[0055] Principle
[0056] The first robot 1a which has actuated joints 12a must implement an anti-collision mechanism for safety, which turns out to be incompatible with the assembly which is by definition a collision.
[0057] The present method proposes not to oppose anti-collision, but to use it in an original and completely counter-intuitive way, which not only allows assembly, but makes it possible to ensure that it takes place correctly, and this without sacrificing safety.
[0058] The idea is to always monitor the force undergone by the actuated joints 12a of the first robot 2a, but from a contact of the first part 2a with the second part 2b, to verify that a variation of the force undergone is in accordance with an expected variation representative of the assembly of the first part 2a) with the second part 2b.
[0059] More precisely, until said contact of the first part 2a with the second part 2b, it is verified in “classic anticollision” that the force undergone is permanently lower than a safety threshold, then when the contact begins (which can be established according to the position of the first robot 1, since we know that this contact is at the said waiting position), we compare the force profile with an expected force profile which makes it possible to ensure that: The observed increase in force is quite normal (due to the fitting) and therefore an impact with an obstacle; and that The assembly is "good" and therefore part 2a+2b is correct, otherwise it means that the assembly went wrong (possibly due to a defect in one of the parts) and the "pseudo" part 2a+2b is rejected.
[0060] Method
[0061] It is possible to start from a state in which the second part is held by the gripping member 13b of the second robot 10b in the predefined waiting position, if the second part 1b is not arranged in bulk, or even if an operator manually gives the second part 2b to the second robot 1b.
[0062] Alternatively, with reference to [Fig. 2], the method begins with optional steps (a) of determining a second trajectory of the third robot allowing it to grasp said second part 2b in the second container 3b gripping member 13c of the third robot 1c and to move it to said waiting position; and (b) of controlling the third robot 1c so as to implement said second trajectory (preferably by estimating a force undergone by said actuated joints 12c of the third robot 2c as a function of the progress of the trajectory).
[0063] The second trajectory of the third robot advantageously comprises it successively: An initial phase, in which the third robot 1e moves from a starting position to a gripping position in which the gripping member 1e of the first robot 1e can grasp the second part 2b (in the second container 3b); and An approach phase, in which the third robot moves it, holding the second part 2b, from the gripping position to the waiting position.
[0064] Step (b) then comprises, where appropriate (if the third robot 1e is different from the second robot 1b, and in particular if it is the first robot 1a as in the examples shown), the transfer of said second part 2b from the gripping member 11e of the third robot 1e to the gripping member 11b of the second robot 1b, as seen in [Fig. 3a]. Said transfer is static: having arrived in the waiting position, the gripping member 11b of the second robot 1b grasps the second part 2b, then the gripping member 11e of the third robot 1b releases the second part 2b (for a brief instant both hold the second part 2b at the same time). The third robot then changes position to free up space in front of the second piece 2b (e.g. return to said starting position).
[0065] Then, the method comprises main steps (c) of determining a first trajectory of the first robot allowing it to grasp said first part 2a in the first container 3a with the gripping member 13a of the first robot 1a and to move it until it is assembled with the second part 2b (see below); and (d) of controlling the first robot 1a so as to implement said second trajectory.
[0066] Like the second trajectory, the first trajectory of the first robot advantageously comprises successively: - An initial phase, in which the first robot 1a moves from a starting position (which can be the waiting position to save time if the third robot 1a is the first robot 1a - one can alternatively return to the starting position) to a gripping position in which the gripping member 1a of the first robot 1a can grasp the first part 2a (in the first container 3a); - An approach phase, in which the first robot 1a moves, holding the first part 2a, from the gripping position to a contact position of the first part 2a with the second part; - A fitting phase of the first part 2a with the second part 2b. The latter includes, visibly or not, two sub-phases of sliding (i.e. start of fitting, reversible, with a resistance which remains limited) and assembly (i.e. end of fitting, irreversible, until the parts are actually assembled and which involves overcoming a strong resistance). These two sub-phases can be done either in one go, i.e. using only the gripping member 11a of the first robot 1a (we speak of direct fitting), if the gripping member 11a of the first robot 1a is sufficiently robust, or be separated if the gripping member 11a is too fragile or if we do not want to risk deregulating it (it must remain very precise).In this second case, of indirect fitting, the sliding sub-phase is implemented with the gripping member 11a of the first robot 1a and the assembly sub-phase is implemented with the support surface 13a of the first robot 1a. More precisely: . • The sliding is to a depositing position in which the gripping member 1a of the first robot 1a can release the first part 2a (represented by figure 3b), in fact it holds on the second part 2a although in an unsecured manner; • The assembly is obtained by applying a thrust to the first part with the support surface 13a of the first robot 1a (represented by Figure 3c). To rephrase, the first robot 1a can move away from the first part 2a and pivot so as to return to its contact not with the gripping member 11a but with a bearing surface 13a (a rigid member generally in the immediate vicinity of the gripping member 11a at the end of the robot arm 1a) which can apply a much higher force.
[0067] Apart from possibly the last fitting phase which has particular constraints which will be described in detail, steps (c) and (d) are respectively similar with the possible steps (a) and (b), and are generally known to those skilled in the art, who may refer in particular to application FR2204180: - step (c) may comprise the prior location of said first part 2a in the first container 3a as a function of at least one three-dimensional point cloud of the parts arranged in bulk in the first container 2a, acquired by the stereoscopic camera 10 (and similarly step (a) may comprise the prior location of said second part 2b in the second container 3b as a function of at least one three-dimensional point cloud of the parts arranged in bulk in the second container 2b, acquired by the stereoscopic camera 10) - the first trajectory (just like the second trajectory) can be determined by a planning algorithm, knowing the required passage positions, and a geometry of the parts 2a, 2b defining a fitting direction; - The execution of a trajectory is done with an adequate controller of the actuated joints 12a, 12b, 12c of the robots 1a, 1b, 1e.
[0068] In a very original step (e) (which can in practice be implemented in real time simultaneously with the executions of trajectories of steps (b) and (d), the data processing means 4 verify that during the trajectory: - Until the first part 2a comes into contact with the second part 2b, the force experienced is less than a safety threshold (for example 30N - this is classic anti-collision) - which corresponds to the entire second trajectory, and the initial and approach phases of the first trajectory; - Then - which corresponds to the fitting phase of the first trajectory - a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part 2a with the second part 2b.
[0069] If: - before the contact of the first part 2a with the second part 2b, the force undergone reaches said safety threshold, the first robot (or the third robot) if we are at step (b), and preferably each robot) is stopped; - after contact of the first part 2a with the second part 2b: • the variation of the force undergone is not in accordance with the expected variation representative of the assembly of the first part 2a with the second part 2b, we can assume that the assembly went wrong and the result of the assembly is rejected (one of the robots 1a, 1b, 1c can put it in the fourth container 3d). As we will see later there can be several cases of rejection. • the variation of the force undergone is in accordance with the expected variation representative of the assembly of the first part 2a with the second part 2b, we can assume that the assembly went well and the result of the assembly (part 2a+2b) is accepted (one of the robots 1a, 1b, 1c can put it in the third container 3c).
[0070] By “contact of the first part 2a with the second part 2b” is meant a theoretical point in time (the contact lasts in practice throughout the fitting phase), which does not need to be detected because: - we know the waiting position (which is predetermined) and the geometry of parts 2a, 2b, so we can determine said contact position; - we know in real time the position of the first robot there, so the processing means 4 can assume that contact takes place when the contact position is reached.
[0071] Thus, as long as the contact position has not been reached, we are in “force experienced is less than a safety threshold” mode and then we move to “variation of the force experienced is in accordance with an expected variation representative of the assembly” mode.
[0072] Mathematically, said force undergone by said actuated joints 12a, 12c as a function of the progress of the trajectory can be a function of the differences, for each actuated joint 12a, 12c, between a predictable reference torque (due to the weight of the robot arm, to inertia, etc.) and a torque actually measured for the joint (it is known how to measure such a torque as a function of the electric current observed in the actuator).
[0073] We can note Ef fortSubi = d{EffortRef,EffortMes), with d a function of distance (typically norm 1, norm 2 or norm infinity), EffortRef^yCoupleRef^ EffortMes^^CoupleMes^ the efforts reference and measured and n the number of joints (6 in the example of the Staübli arm).
[0074] If d is the norm 2 we have E ffortSubi = 12 \CoupleRef. - CoupleMes\ is
[0075]
[0076] ■ / v” Z ; ., Ç and if d is the infinity norm we have EffortSubi = fLi=^TorqueRejTorqueMeas.] Force Suffered = max^TorqueRef.-TorqueM We understand that in the absence of collision, EffortSubi should be almost zero, so that we can predict a low safety threshold, for example 30N, ten or even 100 times lower than the force necessary for fitting (as we will see, we can go up to more than 7500N). By variation of the force undergone, we mean a force profile as a function of the progress of the trajectory, i.e. time or the position of the first robot during the fitting. Generally, the fitting is in a constant direction (translation), so that we can express the force undergone as a function of a fitting dimension according to the fitting direction.
[0077] The verification that the variation of the force undergone is in accordance with an expected variation representative of the assembly of the first part 2a with the second part 2b can be carried out by any technique, including the implementation of an AI classification model, learned on a basis of expected variations representative of the assembly of the first part 2a with the second part 2b (as training examples).
[0078] Alternatively, certain criteria are verified, preferably that: - in the sliding sub-phase, an increase in the force experienced is beyond a minimum threshold (typically at least 20 times the safety threshold) but without reaching an intermediate threshold (typically 3 or 4 times the minimum threshold). Indeed, during sliding the force increases progressively. If the minimum threshold is not reached, it is because the adjustment of the parts is too loose and assembly will be impossible, and if the intermediate threshold is reached, it is because there is already a blockage or an obstacle (remember that the thrust has not yet started with the support surface 13a). We can also check that the force increases substantially linearly during sliding (because there is more and more material involved), for example by checking that the derivative is in a given interval. - In the assembly sub-phase, reaching a maximum threshold (which indicates the arrival of the first part 2a at the stop), higher than the intermediate threshold (and typically between 2 and 3 times the intermediate threshold), at a robot advance position 1a in a predetermined interval. If the force increases too early in the movement: the second part 2b is too long / the bore too short or the first part 2a too tight. If the force increases too late, the second part 2b is too short / the bore too long.
[0079] Mathematically: - For criterion 1, between xc the contact position and xg a predefined end of sliding position (which can be the possible removal position mentioned before):
[0080] 3 xg [xa xg], EffortSubi(x) > ThresholdEffortMin, and Vxg [x^xg], EffortSubi^x} <SeuilEffortIntermédiaire - Pour le critère 2, avec [xmin,xmax] l’intervalle prédéfini de fin d’assemblage au-delà de xg la position prédéfinie de fin de glissement :
[0081] V xg [xg, xmin], Effort Subi (x) < ThresholdEffortMax and 3 xg [xmin, xmax], EffortSubi (x) = ThresholdEf fortMax
[0082] In numerical terms, if the safety threshold is 30N, we can for example provide 1000N for the minimum threshold, 3000N for the intermediate threshold and 7500N for the maximum threshold.
[0083] Figures 4a and 4b represent two examples of force undergone as a function of the fitting dimension, respectively in a case of correct fitting and incorrect fitting. The dimensions are arbitrarily defined, so that they do not correspond in the two figures.
[0084] In [Fig.4a], the position at which the fitting begins corresponds to the 127 mm dimension, and we see the force increasing to 2500N at the 129 mm dimension (end of sliding). The “angular point” observed corresponds to the passage to the support surface 13a. Then, the force continues to increase to 7500N at a dimension between 131 and 132 mm, as expected. The assembly is correct: part 2a+2b will hold and is accepted.
[0085] In [Fig.4b], the position at which the fitting begins corresponds to the dimension at 302 mm, but no increase in force is seen. Between 302 and 304 mm there are jolts which indicate contact but no proper sliding: the axis touches the edge of the bore but this is too large. 1000N is reached only at the dimension 312 mm, then suddenly the threshold 7500N a little before 316 mm. Part 2a+2b will not hold and must be rejected.
[0086] In optional final steps (f) and (g) (the assembled part could simply be dropped by the second robot 2b, or recovered by an operator), the processing means 4 determine a third trajectory of the first third robot allowing it to deposit the assembled part 2a+2b to another container according to the result of the verification of step (e): in our case, the third container 3c if the verification is positive (correct assembly), the fourth container 3d otherwise (incorrect assembly); and finally command the first robot 1a so as to implement said third trajectory.
[0087] As explained, in the case of an incorrect fitting, a notification can be transmitted to interface 6, possibly with an indication of the error:
[0088] E-: force too low during fitting (minimum threshold not reached)
[0089] E+: force too high during fitting (intermediate threshold reached)
[0090] D-: Fitting dimension reached too early (maximum threshold reached before the minimum dimension)
[0091] D+: Fitting dimension reached too late (maximum threshold reached after the minimum dimension)
[0092] System
[0093] According to a second aspect, the invention relates to the assembly system for implementing the method according to the first aspect.
[0094] The system comprises a first robot 1a provided with actuated joints 12a and a gripping member 11a, a second robot 1b different from the first robot 1a and provided with a gripping member 11b, where appropriate a third robot 1e provided with actuated joints 12c and a gripping member 11e (which may be the first or the second robot, or a still distinct robot) and data processing means 4 (advantageously a three-dimensional camera 10, and possibly a memory 5 and an interface 6).
[0095] The data processing means 4 are configured to: - determining a first trajectory of the first robot allowing it to grasp said first part 2a in the first container 3a with the gripping member 11a of the first robot 1a and then to move it until it is assembled with the second part 2b, when said second part 2b is held by the gripping member 11b of the second robot 1b in a predefined waiting position; - Control the first robot 1a so as to implement said trajectory, by estimating a force undergone by said actuated joints 12a of the first robot 2a as a function of the progress of the trajectory; - Check that during the trajectory: • Until contact between the first part 2a and the second part 2b, the force experienced is less than a safety threshold; • then, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part 2a with the second part 2b.
[0096] The data processing means 4 can further be configured to (in the case of a third robot 1e and a second part 2b arranged in bulk): - determining a second trajectory of the third robot allowing it to grasp said second part 2b in the second container 3b with the gripping member 11e of the third robot 1e and to move it to said waiting position; - Command the third robot so as to implement said second trajectory.
[0097] The data processing means 4 can even be configured to: - determining a third trajectory of the first robot allowing it to deposit the result of the assembly to another container 3c, 3d depending on the result of the verification that during the trajectory, after the contact of the first part 2a with the second part 2b, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part 2a with the second part 2b; - Command the first robot to implement said third trajectory.
[0098] Computer program product
[0099] According to a third and a fourth aspect, the invention relates to a computer program product comprising code instructions for the execution (in particular on the data processing means 4) of a method according to the first aspect of assembling at least a first part 2a with a second part 2b, as well as storage means readable by computer equipment (the memory 5) on which this computer program product is found.
Claims
Claims
1. Method for assembling at least a first part (2a) with a second part (2b), the first part (2a) being arranged in bulk in a first container (3a), the method being characterized in that it comprises the implementation by data processing means (4) of steps of: a. Determining a first trajectory of a first robot (1a) provided with actuated joints (12a) and a gripping member (11a), making it possible to grip said first part (2a) in the first container (3a) with the gripping member (11a) of the first robot (1a) and then to move it until it is assembled with the second part (2b), said second part (2b) being held by a gripping member (11b) of a second robot (1b) different from the first robot (1a) in a predefined waiting position; b. Control of the first robot (1a) so as to implement said trajectory, by estimating a force undergone by said actuated joints (12a) of the first robot (2a) as a function of the progress of the trajectory; c. Verification that during the trajectory: - Until contact between the first part (2a) and the second part (2b), the force experienced is less than a safety threshold; - then, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part (2a) with the second part (2b).
2. A method according to claim 1, wherein the second part (2b) is also arranged loose in a second container (3b), the method comprising prior steps of: a. Determination of a second trajectory of a third robot (le) provided with actuated joints (12c) and a gripping member (11c) making it possible to grasp said second part (2b) in the second container (3b) with the gripping member (1le) of the third robot (le) and to move it to said waiting position; b. Control of the third robot (the) so as to implement said second trajectory.
3. Method according to claim 3, in which: - Either the third robot (le) is the second robot (1b); - Or the third robot (le) is not the second robot (1b), preferably the first robot (la), and step (b) comprises the transfer of said second part (2b) from the gripping member (1le) of the third robot (le) to the gripping member (11b) of the second robot (1b).
4. Method according to one of claims 1 to 3, in which step (c) comprises the prior location of said first part (2a) in the first container (3a) as a function of at least one three-dimensional point cloud of the parts arranged in bulk in the first container (2a), acquired by a stereoscopic camera (10).
5. Method according to one of claims 1 to 4, in which the first trajectory of the first robot (1a) successively comprises: - An initial phase, in which the first robot (1a) moves from a starting position to a gripping position in which the gripping member (11a) of the first robot (1a) can grip the first part (2a); - An approach phase, in which the first robot (1a) moves, holding the first part (2a), from the gripping position to the contact position of the first part (2a) with the second part (2b); - A phase of fitting the first part (2a) with the second part (2b).
6. Method according to claim 5, in which the fitting phase comprises a first sliding sub-part a second assembly sub-part, step (e) comprising the verification that: - on the sliding sub-phase, an increase in the force undergone is beyond a minimum threshold but without reaching an intermediate threshold; - On the assembly sub-phase, a maximum threshold greater than intermediate threshold is reached within a predetermined interval of advancement position of the first robot (la).
7. Method according to one of claims 5 and 6, in which the fitting phase comprises a first sub-phase of sliding to a depositing position in which the gripping member (11a) of the first robot (1a) can release the first part (2a), and a second sub-phase of assembly by applying a thrust to the first part with a bearing surface (13a) of the first robot (1a).
8. Method according to one of claims 1 to 7, in which step (e) comprises: - If, before the contact of the first part (2a) with the second part (2b), the force undergone reaches said safety threshold, each robot (1a, 1b, 1c) is stopped; - If, after the contact of the first part (2a) with the second part (2b), the variation of the force undergone is not in accordance with the expected variation representative of the assembly of the first part (2a) with the second part (2b), the result of the assembly is rejected
9. Method according to one of claims 1 to 8, comprising additional steps of: a. Determining a third trajectory of the first robot (la) allowing the result of the assembly to be deposited into another container (3c, 3d) according to the result of the verification of step (e); b. Controlling the first robot (la) so as to implement said third trajectory.
10. System for assembling at least a first part (2a) with a second part (2b), the first part (2a) being arranged loose in a first container (3a), comprising a first robot (1a) provided with actuated joints (12a) and a gripping member (11a), a second robot (1b) different from the first robot (1a) provided with a gripping member (11b), and data processing means (4); characterized in that the data processing means (4) are configured for: - determining a first trajectory of the first robot (1a) making it possible to grasp said first part (2a) in the first container (3a) with the gripping member (11a) of the first robot (1a) and then to move it until it is assembled with the second part (2b), when said second part (2b) is held by the gripping member (11b) of the second robot (1b) in a predefined waiting position; - Controlling the first robot (1a) so as to implement said trajectory, by estimating a force undergone by said actuated joints (12a) of the first robot (2a) as a function of the progress of the trajectory; - Check that during the trajectory: • Until contact between the first part (2a) and the second part (2b), the force experienced is less than a safety threshold; • then, a variation in the force undergone is consistent with an expected variation representative of the assembly of the first part (2a) with the second part (2b).
11. System according to claim 10, comprising a third robot (1e) provided with actuated joints (12c) and a gripping member (11c), the second part (2b) being arranged loose in a second container (3b); the data processing means (4) are configured to: - determining a second trajectory of the third robot (le) making it possible to grasp said second part (2b) in the second container (3b) with the gripping member (11c) of the third robot (le) and to move it to said waiting position; - Command the third robot (the) so as to implement said second trajectory.
12. Computer program product comprising code instructions for executing a method according to one of claims 1 to 9 for assembling at least a first part (2a) with a second
13. part (2b), when said program is executed on a computer. Storage means readable by computer equipment on which is recorded a computer program product comprising code instructions for the execution of a method according to one of claims 1 to 9 of assembling at least a first part (2a) with a second part (2b).
