Landmark2Vec: An Unsupervised Neural Network-Based Landmark Positioning Method
This addresses the problem of mapping landmarks in environments like roads or buildings for agents such as vehicles or mobile devices, but it appears incremental as it builds on unsupervised neural network approaches.
The authors tackled the problem of unsupervised landmark map estimation from agent measurements, introducing Landmark2Vec, a neural network method that learns landmark positions up to scale, rotation, and shift without ground truth data, achieving results in scenarios like visual objects or radio transmitters.
A Neural Network-based method for unsupervised landmarks map estimation from measurements taken from landmarks is introduced. The measurements needed for training the network are the signals observed/received from landmarks by an agent. The definition of landmarks, agent, and the measurements taken by agent from landmarks is rather broad here: landmarks can be visual objects, e.g., poles along a road, with measurements being the size of landmark in a visual sensor mounted on a vehicle (agent), or they can be radio transmitters, e.g., WiFi access points inside a building, with measurements being the Received Signal Strength (RSS) heard from them by a mobile device carried by a person (agent). The goal of the map estimation is then to find the positions of landmarks up to a scale, rotation, and shift (i.e., the topological map of the landmarks). Assuming that there are $L$ landmarks, the measurements will be $L \times 1$ vectors collected over the area. A shallow network then will be trained to learn the map without any ground truth information.