: Harnessing Microbotic Technologies for Pediatric Surgery
Breakthroughs
The surgical tools our Pediabots will use need to be small, cheap, and have extremely high precision. To achieve these breakthroughs, we will need three key technological advances: miniaturized component designs to allow the Pediabots to perform their actions, improved manufacturing techniques to simplify and cheapen creation of Pediabots, and advances in planning algorithms to improve their ability to model information.
Our
Story
Ultrasonic Scalpel
Miniaturized component designs: To invent miniaturized components, we will leverage technologies with properties amenable to miniaturization, such as ultrasonic scalpels. Ultrasonic scalpels vibrate over small ranges with high precision, making them applicable on a small scale.
Improved manufacturing techniques: We will need ways to fabricate the microbots so they have low cost but also high reliability and precision. Many of our components can be made by lithography. Modern lithography machines can be used to make components at the sub-nanometer scale.
Modern Lithography Machine
Modern lithography machine
Our
Story
(define (problem microbot-tumor-problem)
(:domain microbot-tumor)
(:objects tissue1 tissue2 microbot1 tumor1)
(:init (tissue tissue1, tissue2)
(microbot microbot1)
(tumor tumor1)
(at microbot1 tissue1)
(at tumor1 tissue1)
(free microbot1))
(:goal (and (at tumor1 tissue2)
(not (at tumor1 tissue1)))))
Example PDDL code. Press the button below to get the full version.
Advances in planning algorithms: Currently, PDDL does not have many heuristics and designs for medical problems. In order to make our designs practical, we would need breakthroughs involving planning models of surgical problems.
To do this, we will create a research project that will test currently existing models and improve them. We will select a list of existing PDDL planners (e.g., SMTPlan, ENHSP, POPF, etc.) and use sample medical data to test these planners. We will identify which planner can find the most efficient solution. We will also experiment by changing the structure of the input PDDL programs, planning search process and testing such changes to determine optimal configurations (e.g., maximum speed and quality of results).