DEQ-MCL: Discrete-Event Queue-based Monte-Carlo Localization

TL;DR

This work addresses indoor robot self-localization by drawing on hippocampal theta phase precession to formulate a discrete event queue (DEQ)–based Monte-Carlo localization (DEQ-MCL). The method maintains a queue of states across a lag window $L$, forming the posterior $Q(t)=p(x_{t-L:t+L}|a_{1:t},\hat{a}_{t+1:T},o_{1:t},m)$ and integrating past smoothing with current observations while using planned future actions to influence present estimates. Using a particle-filter–style update that samples future states $x_{t+2}$, weights by $p(o_t|x_t,m)$ and $p(x_{t+2}|m)$, and resamples, DEQ-MCL achieves improved RMSE, uncertainty, and variance over traditional MCL baselines in simulated indoor navigation. The brain-inspired framework offers a principled way to fuse memory-like smoothing and foresight from planning into probabilistic self-localization, with potential benefits for real-world indoor robotics and cognitive SLAM research, while motivating future work on neuroscientific validation and links to memory consolidation and active inference.

Abstract

Spatial cognition in hippocampal formation is posited to play a crucial role in the development of self-localization techniques for robots. In this paper, we propose a self-localization approach, DEQ-MCL, based on the discrete event queue hypothesis associated with phase precession within the hippocampal formation. Our method effectively estimates the posterior distribution of states, encompassing both past, present, and future states that are organized as a queue. This approach enables the smoothing of the posterior distribution of past states using current observations and the weighting of the joint distribution by considering the feasibility of future states. Our findings indicate that the proposed method holds promise for augmenting self-localization performance in indoor environments.

Figures (7)

• Figure 1: Phase precession in theta waves and queue representation. The theta phase precession observed in the HF of rodents represents spatial information as a rat moves from left to right. Each field indicates the place receptive field that the rat passes through at that time, and the firing timing of place cells shifts from a late phase to an early phase along with the rat's movement. In the discrete event queue hypothesis, a queue is assigned for each cycle of the theta wave. The variable definition is the same as shown in Fig. \ref{['fig:gm_mcl']}. The variable $x_{3} | o_{1:3}$ indicates the estimated state value at time $t=3$ based on observations up to time $t=3$. Note that the action $a_{t}$ is omitted from the notation.
• Figure 2: Graphical model representation and variable definitions for self-localization. White nodes represent latent variables, while colored nodes represent the observed variables or given parameters.
• Figure 3: Processing of the discrete event queue. In this example, the lag interval of the queue is set to $L=2$. The area enclosed by the thick red border represents the queue that holds the belief distribution of states from $t-2$ to $t+2$ at the current time $t$. The horizontal axis of the table at the bottom represents the time of the estimated state, whereas the vertical axis represents the time of observation (corresponding to the horizontal axis of Fig. \ref{['fig:isousaisa']}). Note that the action $a_{t}$ is omitted.
• Figure 4: DEQ-MCL
• Figure 5: MCL