Search

Search Results

• A METHOD OF CONTROLLING A MOBILE ROBOT USING NATURAL LANGUAGE SEMANTICS

  D.S. Kobzar , V.D. Matveev , Y.D. Lapkin , R.R. Bogdanov , А. S. Izyumov

  2026-04-29

  Abstract ▼

  A large number of different interfaces can be used to control robots, from traditional remotes to augmented reality technologies. However, all such interfaces have a number of limitations, which are particularly acute in service robotics. They are associated with long-term training of a human operator, non-intuitive control for humans, and the need for full human involvement. On the other hand, a new direction has emerged today, related to large language models that are capable of processing natural language and then translating it into robot control commands. There are a number of works demonstrating the possibility of using language models in tasks of planning robot actions. Based on the analysis of existing work, a new method of controlling a mobile robot is proposed, combining the advantages of other methods. The method allows you to plan scenarios for the robot, receiving a natural language mission, the robot's TOP, and information from its sensors. The article also describes the sequence of configuring the system using a large language model to solve this problem. Three variants of instructions for the neural network are presented, which gradually improve the achievability of the generated scenarios. After that, various missions are described, which are set as part of experimental studies - a total of 100 missions were tested, divided into 4 levels of difficulty in equal proportions. The complexity of the missions ranged from describing objects in the robot's field of view to interacting with complex missions involving synonyms of objects and implicitly defined goals. At the end of the work, the results of the evaluation of the algorithm and three variants of the instruction are presented. The conclusion can be considered that the use of language models to assign scenarios to robots is possible, including with a sufficiently high achievement. The model with the most advanced instruction reached 91% of correctly formed scenarios, which suggests the applicability of the developed method for controlling a mobile robot in natural language

• A REVIEW OF METHODS FOR IMPROVING REASONING IN LARGE LANGUAGE MODELS

  V.B. Savinov , N.N. Shusharina

  2026-02-27

  Abstract ▼

  The emergence of large language models has become an important milestone in the field of natural language processing, as such models demonstrate impressive results in text generation, transformation, and analysis, as well as in solving a wide range of applied tasks. However, despite significant practical success, large language models possess limited reasoning capabilities. These limitations manifest in difficulties with generalizing knowledge beyond the training distribution, challenges in transferring knowledge to new contexts, and reduced accuracy when performing multi-step logical and mathematical operations. The goal of this work is to examine methods for improving the reasoning abilities of large language models, where reasoning is understood as the process of forming and evaluating inferences based on existing information. The paper discusses the main types of reasoning relevant to large language models: mathematical, logical, and commonsense reasoning. It provides a list of the most commonly used benchmarks applied to assess the reasoning quality of language models. An overview is presented of the methods used to enhance reasoning in large language models at 2025. Depending on the stage of application (during training or during model usage), the work examines approaches to training data preparation, architectural modifications of language models, training and finetuning procedures (including those using specially constructed synthetic datasets), reinforcement learning, various chain-of-thought construction techniques, mechanisms for integrating external tools, and multi-agent approaches. The paper also discusses existing limitations of large language models, which include the lack of conceptual understanding, poor out-of-distribution generalization, and reduced effectiveness as task complexity increases. Finally, the most promising methods aimed at improving the quality and reliability of reasoning in large language models are highlighted.

• MULTI-PASS METHOD FOR AUTOMATIC CORRECTION OF DISTORTED TEXTS

  D.V. Vakhlakov, V.A. Peresypkin, S.Y. Melnikov

  2021-02-25

  Abstract ▼

  One of the main factors that significantly complicate the understanding, translation and
  analysis of texts obtained by automatic speech recognition or optical recognition of text images
  are the distortions contained in them in the form of erroneous characters, words and phrases.
  The most typical errors of recognition systems are: – replacement of a word with a similar sounding
  or graphic spelling; – replacing several words with one; – replacement of one word with several;
  – skipping words; – insertion or deletion of short words (including prepositions and conjunctions).
  As a result of recognition, a text is obtained that has distortions and consists mainly of dictionary
  words, including in places of distortion. With a large amount of distortion, the texts become
  almost unreadable. Automatic processing of such texts is very difficult, although this task is
  relevant both for Russian and for other common languages. Correction software that works well at
  low distortions in the text, in the case of texts with a high level of distortion, regardless of their
  origin, show unsatisfactory results. This makes it necessary to develop independent approaches to
  correcting distorted texts. A new multi-pass method for correction of distorted texts based on sequential
  error identification and correction of distorted texts is proposed. Non-dictionary word
  forms and word forms which occurrence probability in the text in accordance with the selected
  probabilistic model is less than a preset threshold are considered to be distorted. After setting of
  the distortion sign for individual words, this sign is spread to their combinations, i.e. distorted text
  fragments are extracted. A list of possible word variants which includes only those word forms
  from the dictionary that are located at a certain Levenshtein distance from the word under study is
  built for them. The corrected text from word variants is obtained by searching for the most probable
  chain of word forms. The correction method consists of several passes, at each pass only those
  fragments of the text are corrected that remained distorted after the previous pass of correction.
  The method allows to increase significantly the quality (accuracy) of the correction. In the carried
  out experiments the quality of correction in terms of the F1-measure for moderately distorted texts
  has been increased by 9 %, and for highly distorted texts – by 7.7 %.

• ON THE ACCURACY AND COMPLEXITY OF THE MULTI-STAGE METHOD FOR CORRECTING DISTORTED TEXTS DEPENDING ON THE DEGREE OF DISTORTION

  D.V. Vakhlakov, V. А. Peresypkin, А.V. Germanovich, S.Y. Melnikov, N.N. Copkalo

  130-142

  2021-10-05

  Abstract ▼

  One of the main factors that significantly complicate the understanding, translation and analysis of texts obtained by automatic recognition of speech or images of texts is the presence of distortions in the form of erroneous symbols, words and phrases. Until recently, there were no effective software tools for correcting texts with significant distortions, although this task is rele-vant both for Russian and other common languages in the context of the active use of recognition systems in advanced augmented reality systems. The authors proposed a new multi-stage method for correcting distorted texts, which significantly increases the accuracy of the correction (in terms of the number of correctly corrected words in the text) and is based on the sequential detec-tion of errors and their correction. In this paper, we evaluate the accuracy and computational complexity of the proposed method for correcting distorted texts at various levels of distortion, and determine its place among other modern approaches to correction. The most typical errors of recognition systems are: – replacing a word with a similar sound or graphic spelling; – replacing several words with one; – replacing one word with several; – omission of words; – insertion or deletion of short words (including prepositions and conjunctions). As a result of recognition, a distorted text is obtained, which consists mainly of dictionary words, even in places of distortion. With a large number of distortions, the texts become almost unreadable. Due to the fact that it is problematic to select texts with a wide range of distortion levels in the required amount based on the results of real machine recognition of speech and images of texts, software modeling of distor-tions was used. A text distortion technique has been proposed and implemented that simulates the results of recognition systems in a wide range of distortions; distorted texts have been prepared in the required amount. Within the framework of the proposed multi-stage correction method, non-dictionary word forms and words are considered distorted if the probability of their occurrence in the text in accordance with the chosen language model is less than a given threshold. For such distorted words, a list of possible variants of words is built, which includes only those word forms from the dictionary that are at a certain Levenshtein distance from the word under study. The cor-rected text from the tables of word variants is obtained by searching for the most probable chain of word forms. The correction method consists of several stages, at each stage only those frag-ments of the text that remain distorted after the previous stage are corrected. According to the results of the experiments on the correction of distorted texts, it was concluded that the proposed correction method showed good results with an average value of F-measure >50 % in the distor-tion range from 0 to 75 %. Linguistic experts confirmed the fruitfulness of the proposed approach to correction and its preference over other modern approaches, fixing that with a level of distor-tion of up to 50 % of words, the corrected text is read with much less effort than a distorted one, and with a level of distortion of up to 70% of words, the corrected text also allows you to highlight useful information about the content