This page explains the details of the object detection feature installed on ugo.
Please check the compatibility status by model (manufacture date) to see which models support this feature.
※To use the object detection feature, in addition to this update, an AI model for detection must be remotely applied to the device by our company.
If you wish to use the object detection feature on eligible models such as ugo Pro / Ex (G4) (manufacture date from February to September 2025) or ugo mini (G1) (manufacture date before July 2025), please contact us.
◆Table of Contents
・What is the Object Detection Feature?
・Changes from "Person Detection (Camera)" in "Object Detection"
・Compatibility Status by Model (Manufacture Date)
・Details of the "Object Detection" Command (ugo OS v4.4.0 and later)
・Example Parameter Settings for the "Object Detection" Command
・Example of Creating a Flow Using the Object Detection Feature
◆What is the Object Detection Feature?
The object detection feature is a function where AI detects in real-time whether people are present nearby based on information obtained from sensors such as cameras and LiDAR installed on ugo.
When AI detects a person within the set detection range, this feature automatically activates and proceeds to the next command according to the pre-configured Flow.
◆Changes from "Person Detection (Camera)" in "Object Detection"
|
Traditional "Person Detection (Camera)" Command |
"Object Detection" Command in ugo OS v4.2.0 to before v4.4.0 | "Object Detection" Command in ugo OS v4.4.0 and later | |
| Supported Models | ugo Pro / Ex (manufacture date up to February 2025) | ugo Pro / Ex (G4) (manufacture date from March 2025) mini (G1) |
ugo Pro / Ex (G4) (manufacture date from March 2025) mini (G1) |
| Analysis Speed | Once every 1 second | Once every 0.1 second | Once every 0.1 second |
| Range Specification | Width and depth specified | ・【Pro / Ex / mini】Distance only (circular = circumference from LiDAR) ※For mini, recognition distance may not always be accurately reflected even if specified |
・【Pro / Ex only】Distance range specification (circular = specified based on circumference distance from LiDAR) ・【Pro / Ex / mini】Image range specification (specified based on video on the operation screen) ※Pro / Ex can choose either method for range specification |
◆Compatibility Status by Model (Manufacture Date)
| Model | ugo OS Version | Support Status of "Object Detection" Command | Support Status of Detection Range Specification in "Object Detection" |
| Pro (manufacture date before February 2025) | Below v4.4.0 |
Not usable ※Only the traditional "Person Detection (Camera)" command is available |
– |
| v4.4.0 and later | Usable (with restrictions) |
"Detection distance specification" → Supported only when using front camera or face camera | |
| "Image range specification" → Supported only when using front camera | |||
| Pro (manufacture date from March 2025) | v4.2.0 to before v4.4.0 | Usable | Supported only for distance-based range specification and when using front camera |
| v4.4.0 and later | Usable (with restrictions) |
"Detection distance specification" → Supported only when using front camera or face camera | |
| "Image range specification" → Supported only when using front camera | |||
| mini (G1) | v4.2.0 to before v4.4.0 | Partially usable (with restrictions) |
"Detection distance specification" → Not supported ※Although objects can be detected when using the gimbal 4K camera, the detection distance specification setting cannot be used |
| v4.4.0 and later | Partially usable (with restrictions) |
"Detection distance specification" → Not supported ※Although objects can be detected when using the gimbal 4K camera, the detection distance specification setting cannot be used |
|
| "Image range specification" → Supported only when using the gimbal 4K camera |
◆Details of Detection Modes
The latest "Object Detection" command allows you to select from the following two specification methods according to your purpose.
1. Detection Distance Specification (LiDAR Integration)
This specifies the detection range based on a "circular distance" centered on ugo by utilizing information from the LiDAR sensor.
Advantage: Regardless of the size on the video, AI can determine physically whether a person is within ○ meters from ugo.
Notes:
ugo’s human detection works by combining "AI video recognition" and "LiDAR distance measurement."
AI: Recognizes "people" from the video.
LiDAR: Measures the distance to objects in the direction of the recognized "person."
Due to this mechanism, in environments where there are fixed objects or walls near the front of ugo, if AI recognizes a person far away, LiDAR may mistakenly judge a nearby fixed object as the target, causing a false detection of "a person is nearby" (distance misrecognition). Please be aware of this.
|
|
2. Image Range Specification (Camera Video Analysis)
This specifies the range by the "width and height (percentage position from the bottom center of the operation screen)" of the camera video displayed on the operation screen.
Advantage: Useful when you want to respond only to people standing in the center of the video on the operation screen.
Notes:
①Since the judgment is made based on "position on the screen" rather than actual distance, it cannot distinguish between people far away and those nearby.
②Because the image range specification is based on the bottom center of the operation screen as a reference point, you cannot set it to detect only the upper half or only the right half of the video on the operation screen.
|
|
|
Example of Image Range Specification Settings
|
◆Details of the "Object Detection" Command (ugo OS v4.4.0 and later)
When using the object detection feature in a Flow, use the "Object Detection" command.
It is located under the 【Recognition】 tab in the command picker.
|
|
| Parameter | Description | |
| ① | Recognition Distance |
【The meaning of this setting changes depending on the detection mode】 You can limit the distance between the detection target and ugo. For Detection Distance Specification: This applies only when "Front Camera/Gimbal Camera" or "Face Camera" is selected for "③ Target Camera" in Recognition Distance. Settable range: 0.5[m] to 5[m] For Image Range Specification: This applies only when "Front Camera/Gimbal Camera" is selected for "③ Target Camera" in Recognition Distance. Specify the range upward from the bottom center of the operation screen as 0% (10% to 100% in 1% increments). Settable range: 10[%] to 100[%] |
| ② | Recognition Width |
For Detection Distance Specification: Please leave this blank. If you enter a parameter value, it will automatically be set to "0," but the detection distance specification will still work correctly. ※If you mistakenly enter a parameter value, it will be ignored. For Image Range Specification: This applies only when "Front Camera/Gimbal Camera" is selected for "③ Target Camera" in Recognition Distance. Specify the horizontal range from the bottom center of the operation screen as 0% (10% to 100% in 1% increments). Settable range: 10[%] to 100[%] |
| ③ | Response Time [sec] | This is the time interval over which data is aggregated. Analysis results every 0.1 seconds are summarized and judged over this time. |
| ④ | Target Camera | Select the camera used for detection. |
| ⑤ | Detection Confidence Threshold | Set the confidence score at which AI judges an object as a detection target. Higher settings reduce false detections; lower settings make detection easier. |
| ⑥ | Detection Rate Threshold (%) | Set the percentage of total analyses within the response time that must detect the target to judge it as detected. Higher values can ignore brief crossings. (Example: For a 1-second response time, analysis is done 10 times.) |
| ⑦ | Detection Mode | Select whether to use distance-based or image-based detection. Choose from the following six options: ・"Detect If Present": Proceeds if an object is within recognition distance. ①If "Recognition Distance" is unspecified, LiDAR is not used; the camera's view range is targeted. ・"Detect If Absent": Proceeds if no object is within recognition distance. ①If "Recognition Distance" is unspecified, LiDAR is not used; the camera's view range is targeted. ・"Detect If Present (Distance)": Proceeds if an object is within recognition distance [m]. ・"Detect If Absent (Distance)": Proceeds if no object is within recognition distance [m]. ・"Detect If Present (Image)": Proceeds if an object is within the recognition range [%] on the operation screen. ・"Detect If Absent (Image)": Proceeds if no object is within the recognition range [%] on the operation screen. |
| ⑧ | Detection Target | Currently, only Person is selectable. |
◆Example Parameter Settings for the "Object Detection" Command
①When using in a Flow to detect people passing in front of ugo
Set a short response time to avoid missing brief detections.
The following settings use the Detection Distance Specification feature to detect people passing within 3 meters in front of ugo.
Recognition Distance: 3 (m) (Set according to your operational environment)
Recognition Width: blank (If left blank, "0" is automatically entered, but it works correctly) Response Time (sec): 0.4 (To speak before the person passes by) Target Camera: Front Camera/Gimbal Camera Confidence Threshold: 0.8 Detection Rate Threshold (%): 50 (Consider detected if 2 out of 4 times detected) Detection Mode: Detect If Present (Distance) Detection Target: Person
②When using in a Flow to detect a person standing in front of ugo
By extending the response time to 3 seconds, it is adjusted not to detect people just passing by.
The following settings use the Detection Distance Specification feature to detect people standing within 2 meters in front of ugo.
Recognition Distance: 2 (m) Response Time (sec): 3 Target Camera: Front Camera/Gimbal Camera Confidence Threshold: 0.7 Detection Rate Threshold (%): 60 to 80 (Executes if recognized for about 2 to 2.5 seconds out of 30 times in 3 seconds) Detection Mode: Detect If Present (Distance) Detection Target: Person
◆Example of Creating a Flow Using the Object Detection Feature
We have published an example Flow using the "Object Detection" command (a Flow that greets people passing in front of ugo).
Please refer to the following pages.
・How to Create a Flow to Greet People [ugo Pro]
・How to Create a Flow to Greet People [ugo mini (G1)]