Autonomous Vehicles – Shaoshan Liu, Zhe Zhang, Grace Tsai, Perceptln Shenzhen Ltd
Abstract for “Visual-inertial positioning awareness for autonomous and un-autonomous devices”
“The positional awareness techniques described in this article employ visual-inertial sensor data collection and analysis hardware. References are made to specific examples to illustrate how these improvements can be implemented in the use sensors, techniques, and hardware design. These embodiments can provide machine positional awareness with increased speed and accuracy.
Background for “Visual-inertial positioning awareness for autonomous and un-autonomous devices”
“The discussion of the subject matter in this section is not to be taken to be prior art simply because it has been mentioned in this section. A problem that is mentioned in this section, or related to the subject matter given as background, should not be assumed previously to have been recognized in prior art. This section does not contain any specific approaches. They can, however, correspond to different implementations of the claimed technology.
“Autonomous robots are a long-standing science fiction fantasy. The robot must be able identify their location, determine where they’ve been, and plan where they want to go. This is a technical challenge when creating an autonomous robot. Although SLAM techniques are more reliable and accurate than ever in recent years, it is still difficult to provide robots with fast, accurate, and reliable positional information.
“With the rapid proliferation of virtual reality headsets like the Oculus Rift? and PlayStation?, PlayStation? VR, Samsung Gear? VR, HTC Vive? There has been a rise in demand for a new type of device, one that isn’t autonomous, but can be worn by a human. This would provide accurate, reliable and fast positional information. There are many technical issues that remain in the area of enabling machines and devices, to locate where they are, where their past is and plan where they will go. Recognizing a location and any obstructions quickly and accurately is a particularly difficult area. There have been many different methods. RFID/WiFi methods have been expensive and limited in accuracy. The cost of depth sensor-based methods has been high and they suffer from interference and power drain. Marker-based methods require that markers be placed within the work area. This limits the area where the device can function. Visual approaches are slow and can fail in fast-moving applications. These approaches may also be subject to scale ambiguity. These implementations did not meet the requirements for widespread adoption.
“The problem of providing affordable, reliable and fast positional awareness to devices has remained largely unsolved.”
Summary for “Visual-inertial positioning awareness for autonomous and un-autonomous devices”
“The discussion of the subject matter in this section is not to be taken to be prior art simply because it has been mentioned in this section. A problem that is mentioned in this section, or related to the subject matter given as background, should not be assumed previously to have been recognized in prior art. This section does not contain any specific approaches. They can, however, correspond to different implementations of the claimed technology.
“Autonomous robots are a long-standing science fiction fantasy. The robot must be able identify their location, determine where they’ve been, and plan where they want to go. This is a technical challenge when creating an autonomous robot. Although SLAM techniques are more reliable and accurate than ever in recent years, it is still difficult to provide robots with fast, accurate, and reliable positional information.
“With the rapid proliferation of virtual reality headsets like the Oculus Rift? and PlayStation?, PlayStation? VR, Samsung Gear? VR, HTC Vive? There has been a rise in demand for a new type of device, one that isn’t autonomous, but can be worn by a human. This would provide accurate, reliable and fast positional information. There are many technical issues that remain in the area of enabling machines and devices, to locate where they are, where their past is and plan where they will go. Recognizing a location and any obstructions quickly and accurately is a particularly difficult area. There have been many different methods. RFID/WiFi methods have been expensive and limited in accuracy. The cost of depth sensor-based methods has been high and they suffer from interference and power drain. Marker-based methods require that markers be placed within the work area. This limits the area where the device can function. Visual approaches are slow and can fail in fast-moving applications. These approaches may also be subject to scale ambiguity. These implementations did not meet the requirements for widespread adoption.
“The problem of providing affordable, reliable and fast positional awareness to devices has remained largely unsolved.”
Click here to view the patent on Google Patents.