`
`(19) Japan Patent Office (JP)
`
`Unexamined Patent Application H11-212642
`
`(12) Japanese Unexamined Patent
`Application Publication (A)
`
`
`
`
`(11) Japanese Unexamined Patent
`Application Publication Number
`H11-212642
`
`(43) Publication Date August, 6 1999
`
`(51) Int Cl6 ID Symbols F1
`G05D 1/02 G05D 1/02 L
`A47L 11/10 A47L 11/10
`
`
` Examination Claim Unexamined, Number of Claim Items 15 FD (Total pages 16)
`
`(21) Patent Application No. Patent Application H10-23753
`(22) Application Date January 21, 1998
`(71) Applicant 00000[illeg] 326
`
`
`
`
`Honda Motor Co. Ltd
`
`
`
`
`2-1-1, Miyami-Aoyama, Minatoku, Tokyo
`
`(72) Inventor
`
`
`
`(72) Inventor
`
`
`
`(74) Agent
`
`
`
`
`
`
`
`
`Ichiro Ueno
`c/o Honda R&D, 1-4-1 Chuo, Wako City, Saitama Pref.
`Kato Hironobu
`c/o Honda R&D, 1-4-1 Chuo, Wako City, Saitama Pref.
`Patent Attorney Koki Tanaka (one more)
`
`
`
`
`(54) [Title of Invention] Method and Device for Controlling Self-propelled Robot
`(57) [Abstract] (Revised)
`[Challenge] It is designed for a robot traveling trajectory to fill up the entire region efficiently
`and exhaustively, based on the boundary detection signals of the travel planned region.
`
`[Solution Means] By suitably combing the spiral travel (case a) that rotates and travels while the
`rotation radius from the position where robot 1 is located is made gradually larger, a boundary
`(border)-following travel that travels for a planned time along the boundary, and a random travel
`(case b) that rotates a planned angle in response to detecting the boundary, and after that, travels
`straight, it is executed as in the case (c). A combination sequence can be stored in a memory in
`advance.
`[See Fig. 6 on page 11]
`
`-2-
`
`[Claims]
`[Claim 1] A control method of a self-propelled robot wherein a sensor to detect the boundary of
`a travel planned region is provided and the method allows the robot to travel to fill up the
`aforementioned travel planned region as exhaustively as possible,
`The method for controlling a self-propelled robot wherein a rotation travel is started from an
`optional position in the aforementioned region until the aforementioned boundary (border) is
`
`Silver Star Exhibit 1005
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`detected by the aforementioned sensor, and a spiral travel makes the rotation radius gradually
`larger; this spiral travel, and a border-following travel that travels along the aforementioned
`boundary are alternatively executed.
`
`[Claim 2] The method for controlling a self-propelled robot according to claim 1 wherein when a
`boundary is detected during spiral travel, the aforementioned spiral travel is stopped and it is
`changed to a border-following travel.
`
`[Claim 3] The method for controlling a self-propelled robot according to claim 1 wherein when a
`boundary is detected by the aforementioned sensor during spiral travel, the aforementioned spiral
`travel is stopped, and after repeating the random travel a planned number of times that includes
`the planned angle rotation in response to a boundary detection and a planned distance forward
`advance that follows this, the rotation traveling is executed.
`
`[Claim 4] The method for controlling a self-propelled robot according to claim 1 wherein when a
`boundary is detected during spiral travel, it stops temporarily, and the planned angle rotation and
`forward advance until the boundary is again detected is repeated N times (N is an optional
`integer), and the border-following travel that travels along the boundary that was detected last is
`executed.
`
`[Claim 5] The method for controlling the self-propelled robot according to claim 3 or 4 wherein
`before executing the rotation when the boundary was detected, a retreat of a planned distance is
`executed.
`
`[Claim 6] The method for controlling a self-propelled robot according to any of the claim 1
`through 5 wherein when a traveling is started, a spiral travel mode is executed
`
`[Claim 7] The method for controlling a self-propelled robot wherein a sensor to detect the
`boundary of a travel planned region is provided and the method allows travelling to fill up the
`aforementioned travel planned region as exhaustively as possible,
`the method for controlling a self-propelled robot wherein the rotation traveling is started from a
`certain position in the aforementioned region until the aforementioned boundary is detected by
`the aforementioned sensor, this spiral travel mode makes the rotation radius gradually larger;
`the method comprises the spiral travel mode, the border-following travel mode that travels along
`the aforementioned boundary for a planned time, and a random travel mode wherein when the
`boundary is detected by aforementioned sensor, the aforementioned robot traveling is stopped,
`and the planned angle rotation in response to the boundary detection and planned distance
`forward advance that follows this are executed,
`One of the aforementioned any 3 modes is selected and executed sequentially and at the time,
`before or after the random travel, at least one of spiral travel mode or border-following travel
`mode is executed.
`
`[Claim 8] The method for controlling a self-propelled robot according to claim 7 wherein at the
`time to start traveling, the spiral travel mode is executed
`
`
`Silver Star Exhibit 1005 - 2
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`[Claim 9] The method for controlling a self-propelled robot according to any of the claims 1
`through 7 wherein the sequence of executing spiral travel, border-following travel and random
`travel is preset before traveling start.
`
`[Claim 10] The method for controlling a self-propelled robot according to claim 8 or 9 wherein
`the aforementioned spiral travel mode, random travel mode, border-following travel mode and
`random travel mode is repeated in this sequence.
`
`[Claim 11] The method for controlling a self-propelled robot according to any of the claims 1
`through 10 wherein regarding the border-following travel, based on the boundary detection
`signal positioned on the side of robot main body, when the aforementioned boundary is detected,
`it advances straight, and when the aforementioned boundary is not detected, it is rotated to
`approach the boundary, and when it contacted the one side boundary or get too close, it is rotated
`to get away from the boundary.
`
`[Claim 12] The method for controlling a self-propelled robot according to any of the claims 3
`through 11 wherein the aforementioned rotation angle is somewhat 135 ͦ in the progress
`direction.
`
`[Claim 13] The method for controlling a self-propelled robot according to any of the claims 3
`through 12 wherein each continued time of the aforementioned border-following travel is preset.
`
`[Claim 14] A device for controlling a self-propelled robot wherein the robot travels to fill up the
`travel planned region as exhaustively as possible,
`the device for controlling a self-propelled robot wherein the device is provided with a plural
`number of sensors that are positioned at least in front of the robot main body and on one side
`thereof, and detects that the aforementioned robot approaches within the planned distance from
`the boundary of the aforementioned travel planned region and generates a proximity output;
`a sensor that is positioned at the rim of the robot main body and generates a contact output when
`the aforementioned robot contacted the boundary of the aforementioned travel planned region;
`execution mode setup means that sequentially selects and sets up the traveling mode that the
`robot should execute from among random travel, spiral travel and border-following travel
`mode;
`Control means that controls the robot traveling according to the traveling mode selected and set
`up.
`[Claim 15] The device for controlling a self-propelled robot according to claim 14 wherein the
`aforementioned execution mode setup means comprises the means that pre-stores the traveling
`mode that the robot should sequentially execute, and the means that reads from aforementioned
`memory means the travel mode the robot should execute next in response to the progress of
`traveling mode, and the aforementioned control means controls the robot traveling according to
`the traveling mode that was read.
`
`[Detailed Explanation of Invention]
`[0001]
`[Technical Field of Invention] The present invention relates to a method and device for
`controlling a self-propelled robot, particularly relates to a method and device for controlling a
`
`Silver Star Exhibit 1005 - 3
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`self-propelled robot that can travel the given region as in a short time and exhaustively as
`possible.
`
`[0002]
`[Prior Art] Self-propelled robots such as a sweeping robot, a lawn mowing robot, a field grade
`robot, and an agricultural dispersion robot etc are known that automatically travel a given region
`and executes pre-set work.
`
`-3-
`
`For instance, regarding the sweeping robot described in Japanese Unexamined Patent
`Application H5-46246 Gazette, before starting cleanup, the robot circles inside the room, and
`detects the room size, shape and obstacles and does mapping of the traveling region, that is, the
`mapping of the cleanup region. After that, based on the coordinate information that was obtained
`by this mapping operation, the robot engages in a zigzag traveling and spiral traveling in which a
`circling travel radius is made smaller for each round, and the entire room is cleaned. This robot
`detects the wall surface by a contact sensor and ultrasonic sensor and decides the progress path,
`and also detects the finish of the circling by distance meter. Likewise, a robot that travels
`exhaustively the entire floor surface is also disclosed in Japanese Unexamined Patent
`Application H5-257533 Gazette.
`
`[0003] Regarding the conventional robot described above, various drive system actuators such as
`a motor is controlled so that, based on the information detected by many sensors, traveling
`region conditions are sufficiently grasped and the robot fill ups and travels the travel region
`precisely and efficiently. Because of this, the control system gets quite complicated and high
`priced, and also, the processing speed slows down. Furthermore, there were problems such that
`due to mapping, teaching and various processing, it took a long time and training for initial
`settings such as threshold value settings etc, and obstacle avoidance operations were delayed etc.
`
`[0004] The inventors involved herein previously proposed a method and device for controlling a
`robot (Japanese Unexamined Patent Application H9-29768) in which, regarding cleaning, lawn
`mowing robots etc, targeting the points in some cases in which it is not necessary to travel
`without missing the entire region of the target with high precision, and even if some unworked
`region remained, no big difficulty is generated, the robot can travels the given region somewhat
`exhaustively with a simpler configuration.
`
`[0005] The aforementioned proposed self-propelled robot is equipped with various sensors that
`detect the work region boundary and obstacles, a wheel rotation number sensor etc, and it has
`the spiral travel mode (Fig. 6 a, c) in which centered on the optional point inside the
`aforementioned region, the rotation radius is gradually made larger, and a random travel mode
`(Fig. 6 b) in which when the distance to the boundary or obstacle gets to be within the preset
`values, a spiral travel is stopped and the robot rotates with a preset angle and advances straight
`so that it gets far away from the aforementioned region boundary, and thereafter furthermore,
`every time the aforementioned region boundary is detected, repeats the rotation and straight
`advance a preset number of times (fine tuning). In this case, it was found out as a result of
`simulations that in order to improve the efficiency (hereinafter called [work efficiency] that
`
`Silver Star Exhibit 1005 - 4
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`enables a robot to travel in the region exhaustively and faster, the optimum rotation angle α is
`135 ͦ. Here, the traveling pattern that sets a rotation angle α to be 135 ͦ is called a fine tuning
`random travel.
`
`[0006] During the operation, as shown in Fig. 6 (a) through (c), after doing spiral traveling, it
`moves to the random travel mode, and at the position of a planned distance straight advance from
`the last rotation, it starts the aforementioned spiral travel. The planned number of times of the
`aforementioned rotation and the last straight advance distance are predetermined by a simulation
`model so that the time to attain the desired coverage rate is minimal.
`
`[0007] Fig. 16 is a block drawing showing a hardware configuration of the aforementioned self-
`propelled robot control device. The control devicen7 is equipped with CPU 8 and a drive circuit
`16 controls input and output of an ultrasonic sensor 6. Based on the information from a pair of
`multiple ultrasonic sensors 6 positioned oriented toward front, right and left side surfaces and
`slanting -front direction etc, contact sensor 5A positioned on front end bumper etc, rotation
`number sensor 10 of right and left wheels, CPU 8 controls the operations of right and left wheel
`drive motors 14, 15, right and left brakes 12, 13 etc, enabling the robot to execute each operation
`of moving forward, retreat, stopping and ultra-pivot turn, pivot turn, rapid turn and slow turn.
`Slow turn and rapid turn are executed by making the rotation speed of right and left wheels
`different. As is evident, the rotation radius is decided by the left and right wheel rotation speed
`and its difference. A super pivot turn is a turn executed by making the left and right wheels
`mutually reverse-rotate, and a pivotal turn is a rotation such that one side of left and right wheels
`is stopped and only one side is rotated. The rotation angel in these cases is decided by the
`rotation amount of the wheel to be turned.
`
`[0008] Regarding this robot, it is not that the action plan generated by each sensor status is
`immediately executed but based on the preset urgency degree, it is prioritized and the action plan
`with higher urgency is designed to be executed preemptively.
`
`[0009] Fig. 17 is a block drawing showing the function of action decisions executed by the
`aforementioned robot. In case the action plan AP1, AP2, …, APn are generated based on the
`distance to the obstacle detected by each sensor 6, 5A, selection function 20 selects the action
`plan among action plans AP1 through AP n, that has the highest urgency operation when
`avoiding the collision with the wall surface, and energizes an actuator 19. According to this
`conventional example, when the retreat control is activated, this is regarded as the highest
`urgency operation and is set to have the first priority. Following this, the ultra-pivot turn control
`became the 2nd priority, and after that, the priority was given in order of pivot turn, rapid turn,
`slow turn. Moreover, prioritizing the action plan described above was decided according to the
`distance to the obstacle calculated based on the detection result by ultrasonic sensor 6, and the
`stop control when an obstacle was detected by aforementioned contact sensor 5A is not included.
`
`[0010] Fig. 18 is a graph showing the result of a simulation of work time and work progress
`degree by the robot described above, and Y axis shows the ratio of the region area filled up by
`robot traveling in the given region and X axis shows the elapsed time from the traveling start.
`The flat area of the robot is represented by a circle of diameter 20 cm and the traveling speed is
`
`Silver Star Exhibit 1005 - 5
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`set to be 13cm/ sec. Travel region, in case of Fig (a), is 4.2m x 4.2m square and in case of Fig.
`(b), 4.2m x 8.4m rectangle.
`
`[0011] Moreover, the coordinate system travel stated in the same drawing is a method to travel
`along the course that is preset so that a robot travels covering the work region.
`
`-4-
`
`And according to the said travel method, in linear proportion to the time passage, the covered
`region ratio increases. Compared with this, according to the other travel method that includes a
`spiral travel, since worked area growth is decreased, it is difficult to aim the complete coverage
`of the region. Hence, as an example, when an efficiency comparison is made using the time
`required to travel by covering 80% of the region, in case of Fig. 18(a), in the three travel
`methods that exclude the coordinate system travel, as shown in Fig. 6 (a) through (c)), one can
`tell that the spiral travels combined with fine tuning random travel has the shortest time (about
`1,800 sec) and convers 80% of the region. Moreover, in case of Fig. 18 (b) where the area is
`expanded double, a somewhat similar trend was obtained. Moreover, in this case, the number of
`times for turns to maximize the work efficiency that show the average coverage of percentage of
`the entire travel region per unit time (1 sec) is 5 times, and the straight advance time after turn is
`15 to 30 secs, and it was found out by the aforementioned simulation result that aforementioned
`time and turn number of times did not mutually impact the other.
`
`[0012]
`[Challenge the Invention Attempts to Solve] Even the aforementioned proposed robot can work
`comparatively efficiently up to a degree of coverage or fill up (about 80% of the entire area) but
`when it is tried to increase the covered area more than that, it starts to spend a very long time, the
`challenge is that if for instance a plural number of rooms partitioned by walls etc, and a room
`with furniture are continuously cleaned, work efficiency tends to go down.
`
`[0013] The present invention aims to provide a method and device for controlling a self-
`propelled robot in which it is comparatively easy to increase the covered rate to 80% or more and
`even in case there are obstacles such as partitions and furniture in the work region, the
`aforementioned work region is continuously enabled to be worked on and it is difficult to reduce
`the work efficiency.
`
`[0014]
`[Means to Solve the Challenge] It is designed such that it is equipped with a sensor that detects
`the boundary of travel planned regions and starting a rotation travel from the optional position
`within the aforementioned region, while detecting the aforementioned boundary and obstacles
`by the aforementioned sensor, a spiral travel makes the turn radius gradually larger; the spiral
`travel and a border-following travel that travels along the aforementioned boundary,
`furthermore a random travel are combined, depending on the desire, thereby the aforementioned
`travel planned region is designed to be filled up as exhaustively as possible. When a boundary is
`detected during a spiral travel, the aforementioned spiral travel is stopped and it moves to a
`random travel and a border-following travel. Travel modes can be suitably combined, but in the
`simulation where spiral travel – random travel – border-following - random travel combination
`
`Silver Star Exhibit 1005 - 6
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`was repeatedly executed, the travel planned area is 35m2 or 57m2, and inside the target region
`where obstacles are scattered can be somewhat filled up 100% at the time of 124 minutes and
`271 minutes respectively.
`
`[0015]
`[Embodiments of Invention] The following explains the present invention referring to drawings.
`Fig. 2 is an outline plane view of a self-propelled robot involving an embodiment of the present
`invention. Fig. 3 is an outline side view drawing. In those figures, robot 1 is configured to be
`positioned on both left and right sides of the main body 2 respectively, and to execute each
`operation of forward movement, retreat and stop and rotation by wheels 3, 4 driven by (with an
`unlimited trajectory or simple) separate motor (not shown in figure). In the aforementioned
`wheels 3, 4 are provided sensors (not shown in figures) for detecting each number of rotation.
`Moreover, in the following explanation, in case all the sensors are collectively called, it is called
`simply [sensors 26]. Main body 2 is made of a flexible material and configured somewhat in a
`half-cut egg shell shape, and between its inner circumference and its inner main frame is attached
`a contact sensor (not shown in figure) that detects the contact with an obstacle.
`
`[0016] Furthermore, the robot 1 is provided with a pair of a plural number of infrared sensors on
`left and right symmetrically to detect boundaries and obstacles without contact. That is, in the
`advancing direction in front of the robot 1 is positioned sensors 26R, 26L; in the slanted front
`direction, 26 MR, 26 ML; and in the rear direction, 26 RR ,26RL respectively, and furthermore
`in the left direction, side sensor 25L for a border-following travel unique for the present
`invention is positioned. The added letter R in the aforementioned each symbol is for a right-side
`obstacle detection with respect to the travel direction; and the added letter L is for a left side
`obstacle detection with respect to the travel direction.
`
`[0017] Moreover, not shown in a figure, a side sensor can be provided on the right side of the
`main body. It is desired that these sensors be an infrared sensor, but if it is a proximity sensor
`that can detect the obstacle within a planned short distance (for instance, 10 to 15 cm), any form
`of sensors such as ultrasonic or other optical sensor etc can be used. Regarding the configuration
`of the main body of the aforementioned self-propelled robot and the detail of contact sensors, it
`is described in detail in the patent application (A97-467, 468, filed on December 22, 1997) of the
`separate case by the applicant involved herein, hence the description of the specification is
`quoted and incorporated herein.
`
`[0018] Fig. 1 is a block drawing showing a hardware configuration of the device for controlling
`a self-propelled robot of an embodiment by the present invention, and shows the same symbols
`or equivalent parts as in Fig. 16 and Fig. 2 and 3. As clarified by the comparison with Fig. 16,
`in Fig. 1 the ultrasonic sensor 6 of Fig. 16 is replaced by proximity sensors 25L and 26 such as
`an infrared sensor etc, and the signal of these sensors 25L and 26, and the detection signal of a
`rotation number sensor 10 (encoder) of the motor that drives a contact sensor 5A and left and
`right wheels 3, 4 are inputted into CPU 8 via digital input unit 9.
`
`[0019] On the other hand, CPU 8 is connected via digital output unit 11 with a right wheel
`electromagnetic brake 12, a left wheel electromagnetic brake 13, a right wheel motor 14 (herein
`
`Silver Star Exhibit 1005 - 7
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`after called a right motor), and a left wheel motor 15 (hereinafter called a left motor). Then,
`various instructions based on the processing by CPU 8 are via the digital output unit 11,
`
`-5-
`
`inputted into left and right wheel electromagnetic brakes 12, 13 and left and right motors 14, 15
`etc respectively. Rotation direction instruction signal is supplied to the left and right motors 14,
`15 through digital output unit. Moreover, into the left and right wheel drive motors 14, 15 are
`inputted rotation speed instructions from CPU 8.
`
`[0020] Due to the configuration described above, based on the proximity and contact information
`from sensors 25L, 26 and a contact sensor 5A (hereinafter called [sensors] collectively), CPU 8
`decides the drive system operations of left and right wheel drive motors 14, 15 etc. The said
`robot executes each operation of a straight advancing, a retreat, a stop and a rotation as described
`above, but the control functions for those are realized by the function of CPU 8 separately as a
`module. Information input processing and operation decision processing from each sensor is
`operating all the time, but each control module of an ultra-pivot turn, a stop and a retreat is in a
`sleep state normally, and only the straight advance control is activated. Moreover, as understood
`easily, the rotations other than ultra-pivot turns are included in the straight advance control
`module function.
`
`[0021] The operation decision unit 18 of CPU 8 is configured such that the preset operation
`based on the information from each sensor is executed conditioned reflex- like. As described
`regarding Fig. 17, the operation decision unit 18 is configured to be a hierarchy type that
`corresponds with each sensor, and generates an action plan depending on the signal status from
`sensor 25L, 26, 5A and outputs an execution request. Based on this execution request, drive
`system 19 (actuator) that comprises the left and right wheel electromagnetic brakes and left and
`right motors is controlled. Thus, the action plans that were individually generated based on
`information from each sensor are piled up and the entire robot operations, that is, the operations
`such as a straight advance, a retreat, a stop and a slow rotation, a fast rotation and a pivot turn, an
`ultra-pivot turn etc are decided.
`
`[0022] Moreover, in the embodiment of the present invention also, as described above, it is not
`that the action plan that was generated based on the output from each sensor is immediately
`executed, but that based on the preset urgency degree, they are prioritized and the action plan
`with a higher urgency is preemptively executed. This priority is the same as the robot proposed
`by the inventors involved herein earlier, and excluding the stop control at the time of obstacle
`detections by aforementioned contract sensor, the sequence is a retreat, an ultra-pivot turn, a
`pivot turn, a fast rotation and a slow rotation.
`
`[0023] The characteristics of a robot travel pattern in the embodiment of the present invention
`is, in addition to the aforementioned random travel, fine tuning random travel, spiral travel
`pattern, is the point to have a [border-following travel] that travels along a boundary such as
`walls etc (sometimes called [corner travel]. The border-following travel pattern, while executing
`(fine tuning) random travel and spiral travel patterns, is started when the side sensor 25L detects
`the boundary such as a wall etc, and it is continued for a planned time from then.
`
`Silver Star Exhibit 1005 - 8
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`
`[0024] Fig. 4 is a flow chart that shows border-following travel processing. As described later,
`during random travel and spiral travel, if the side sensor 25L or 26 senses a boundary such as a
`wall, and generates an output, CPU 8 generates a border-following travel start instruction and the
`processing in Fig. 4 is started (Step S70). At Step S71, it makes a straight advance, and at Step
`S72, it decides whether or not the aforementioned side sensor still detects the boundary.
`Moreover, the side sensor detection range decides how much distance the robot can keep from
`the boundary to travel along the boundary, hence it is better if it is not too big and for instance 10
`cm to 15 cm is suitable.
`
`[0025] If the side sensor is no longer detecting the boundary, it is going away from the boundary,
`hence at Step S73, it makes a planned angle slow rotation to approach the boundary, and returns
`to the Step S71 and continues to make a straight advance. At Step S72, if the side sensor detects
`the boundary, then it is traveling along this near the boundary, hence at Step S74, it furthermore
`continues the straight advance. At Step S75, it is decided if the front-end contact sensor 5A
`detects the boundary such as a wall, and if the decision is negative, steps S72 through 75 are
`repeated. On the other hand, if the decision of Step S75 is positive, then at Step S76, it retreats
`for a planned distance and furthermore, it makes a planned angle rotation in the opposite
`direction with the detected boundary, and returns to Step S71 and makes a straight advance.
`Using such a method, the robot of the present invention continues to travel along the boundary
`such as walls etc. The aforementioned border-following travel is stopped after continuing for a
`planned time (or distance), it moves to a random travel mode. The aforementioned planned time
`can be realized by stopping the border-following travel by a suitable timer interrupt, but at Step
`S70, the stop timer is activated and it can be stopped by deciding the aforementioned timer count
`up at Step S71A, 74A shown by the dotted line in Fig.4.
`
`[0026] Continuing, each travel pattern of the robot by the present invention combined with the
`border-following travel described above is explained. First, the random travel which is a basic
`travel pattern of robot 1 is explained. In random travel, as shown in Fig. 5, if the robot 1
`positioned in the region A surrounded by a boundary or a wall surface B makes a straight
`advance and enters within the planned distance from wall surface B, then it makes a return
`operation by doing a temporary stop and a planned angle rotation (depending on the needs, it can
`retreat a planned distance before it), then again makes a straight advance and goes toward a
`different wall surface B. At this time, as to the rotation angle α for a return operation near the
`wall surface B (refer to Fig. 5 (b)), it can be selected at random for every return operation and
`set.
`
`[0027] The inventors involved herein found out that if the aforementioned border-following
`travel is furthermore combined with the spiral travel and random travel pattern in Fig. 6, the
`operation efficiency is furthermore improved in which by combining the spiral travel with a
`random travel, at the point when random travel (fine tuning random travel as much as possible)
`is repeated a planned number of times, a spiral travel is executed)
`
`[0028] Here, spiral travel and random travel are explained furthermore in detail. In Fig. 6, a
`robot 1 is placed inside the region A. This region A assumes a rectangle room surrounded by a
`wall surface B.
`
`Silver Star Exhibit 1005 - 9
`
`

`

`
`
`Unexamined Patent Application H11-212642
`
`
`-6-
`
`The position to place a robot 1 initially is optional. As shown in Fig. 6 (a), the robot 1 starts to
`do spiral traveling at the position where it is placed. The spiral travel is a travel pattern in
`rotation travel in which a rotation radius is gradually made larger by a planned amount, and as
`described in detail later referring to Fig. 10, it is controlled based on an operation decision
`different from a straight advance, an ultra-pivot turn, a retreat etc. Here, in order not to make
`space in a travel trajectory, the speed of left and right wheels 3, 4, that is, the rotation speed of
`each wheel drive motor 14, 15 is calculated and by updating these speeds, the rotation radius is
`gradually increased. A spiral gets bigger and based on the output of sensors 26 and 25L, when it
`is recognized that the robot 1 approached within the planned distance with respect to the wall
`surface B, the spiral travel is stopped and a random travel is started to move to the next spiral
`travel start position (preferably fine-tuning travel) (Fig. 6b). The shadowed part in Fig. 6 (b)
`and (c) is the travel trajectory of robot 1, and that is the region completely filled up by traveling.
`
`[0029] The chance to stop a spiral travel and move to the start position for next spiral travel is as
`follows. Robot 1 approaches the wall surface B, and when it is detected by sensors 26 and 25L
`that the wall B is within the planned distance from the robot, a turn back operation explained in
`Fig. 5 is executed. For instance, if the robot 1 detects the wall surface B, it stops at the position,
`and depending on the needs, after retreating a planned distance, makes a 135 ͦ (or, another
`optional angle) ultra-pivot turn, and turn back and makes a straight advance to get far away from
`the wall surface B. In this case, of course, it can make a pivot turn or a rapid rotation with a
`small angle to avoid the wall surface B.
`
`[0030] In this way, at wall surface B, it turn