Machine-Learning Based Optimization of Navigation Algorithms and embedded implementation techniques for Service Robotics

PhD program in Electrical, Electronics and Communications Engineering


Marcello Chiaberge –

Fabrizio Lamberti

PhD Student: Francesco Salvetti

Context of the research activity

Service robotics is expected to grow significantly in the next few years, fueled by the large number of innovations in the field and the decreasing trend in the cost of the hardware. However, it is not yet completely clear how to use such robots, which include drones, rovers, etc., in the most effective way to create innovative services. Furthermore, the problem may become even harder when a multitude of coordinated objects (e.g., swarm) are required.
In this context, the activity proposed in this PhD proposal will address the navigation optimization, data processing, storage and communication issues that may arise in the service robotics scenario, when robots are equipped with numerous devices, ranging from acquisition sensors to communication ones able to operate on different wireless channels using several access technologies. Typically, such scenario requires to handle a huge amount of information that must be either processed locally or transmitted for processing and storage in the cloud. The scenario poses many challenges in terms of optimizing resource consumption, which include problems such as where to save data, how to communicate with neighbor nodes, how to encode information, how to cope with remote services (e.g., in cloud) that might not always be available due to, for instance, network connectivity issues, and how to optimize the robot navigation algorithms while explicitly considering connectivity requirements and environmental constraints. Wireless coverage, in fact, is highly variable in the location and time dimension and sometimes even not available. This fact has to be explicitly modelled and taken into consideration during the planning and execution of robot movements and trajectories if a given performance, in terms of bandwidth and latency, is needed, and must be coped with real-time navigation and obstacle avoidance issues.
Research in this area is ongoing. For instance, while many efforts have been devoted to designing optimal navigation strategies when the environment and constraints are known, optimizing strategies in real-time on board of the devices is still difficult when the environment is not well known or when any type of connection is not available (e.g. tunnels). The most promising approaches are based on artificial intelligence techniques, including supervised and reinforcement learning, to learn an optimal navigation strategy, in the presence of uncertainty, while at the same time attempting to fulfill all the communication and hardware constraints (e.g., communication bandwidth, latency, storage requirements). The navigation strategy exploits data gathered by the onboard sensors, as well as by other nearby robots or sensors in the same area.
Another challenging aspect is how to automatically perform optimal adaptation of both the path and data to be transmitted, in a way that it is jointly optimized, and satisfies both behavioral as well as computational requirements. Moreover, optimal storage strategies for huge amounts of data need to be sought, since the robots are expected to collect as much information as possible with the onboard sensors; collected data can be processed to allow, for instance, extracting relevant information in a short time frame that can be then re-used for many tasks, avoiding or reducing the need to transmit and  process them in external much more powerful computing system (cloud). Machine learning can again be used to optimize the acquisition, encoding, transmission and storage of the most relevant information for the planning task.
In this PhD proposal, among the many service robotics application fields, a few will be identified, analyzed in detail and addressed, in coordination with the activity carried out by the Interdepartmental Center “Pic4Ser” and “SmartData”. As a starting point, applications will include, but will not be limited to, the so-called Precision Agriculture, where efficient coordination and communication between robots (aerial and ground vehicles) and cloud services is required. Moreover, it is also important to perform smart and targeted data acquisition activities in an autonomous or semi-autonomous fashion and timely communicate results to control and supervision operators or centers.


The main objective of the PhD activities will be the development of algorithms, technologies and systems that can effectively jointly optimize the performance of the considered systems in terms of traditional communication metrics (e.g., bandwidth, latency) and application-layer metrics (e.g., utility in terms of lowest cost navigation paths) which will be defined on a case-by-case basis depending on the application, while at the same time minimizing all costs (e.g., computational complexity, storage cost, number of required devices and their economic cost).
Such objectives will be achieved by using both theoretical and practical approaches. In fact, the adopted methodologies will include the development, wherever possible, of theoretical frameworks to model the whole system, in order to investigate the problem from an analytical point of view. The proposed techniques will build upon recent advances in artificial intelligence, exploiting machine learning to determine optimal planning strategies in the presence of multiple competing constraints and under the presence of uncertainty. Artificial intelligence techniques will be employed both for sensing the environment and encoding semantic (e.g., using supervised learning strategies), and for planning the optimal trajectory (e.g. through reinforcement learning). The research project will leverage resources, data and know-how acquired by the Pic4Ser center for training the intelligent components, as well as define simulation environments in which the artificial intelligence system can be safely and effectively trained. 
The resulting insights will then be validated in practical cases by analyzing the performance of the system with simulations and real-world experiments. Both proponent research groups have extensive expertise in such fields.
In this regard, cooperation with companies will also be sought in order to facilitate the migration of the developed algorithms and technologies to prototypes that can then be effectively tested in real application scenarios.

Skills and competencies for the development of the activity

  • Solid knowledge in Electronics and Computer Engineering
  • Basic knowledge about Computer Networks, Cloud computing, Databases
  • Application Layer analytical development skills
  • Solid knowledge about machine learning (supervised learning, reinforcement learning)
  • Experience with deployment on embedded systems
  • Programming skills
  • Team working capacities

Further information about the PhD program at Politecnico can be found here

Back to the list of PhD positions