Opening Keynote

Zied Bahrouni - CEO @Motius

RISC-V & FPGAs

Sebastian Plamauer - Team Lead Hardware Cluster @Motius

FPGAs (Field Programmable Gate Arrays) have been around for a while now, but their relevance has increased lately. Tech companies are investing heavily to use them in new areas. We want to explore how to use FPGAs in different ways and how to implement custom hardware on them using RISC-V for general computing and interfaces.

Motius Goes Serverless

Sagar Ramachandrappa - Fullstack Developer @Motius

Serverless is a cloud-native development model that allows developers to build and run applications without having to manage servers. This would allow us to work even more efficient in R&D projects.

Event Cameras

Katrine Maria Nielsen - Tech Specialist Robotics @Motius

Event Cameras could revolutionize the way we obtain visual information. The technology is based on bio-inspired vision sensors that output pixel-level brightness changes instead of standard intensity frames. This offers significant advantages over standard cameras, namely a very high dynamic range, no motion blur, and a latency in the order of microseconds. There could be a variety of use cases e.g. ML algorithms for processing, Quantum physics applications (where speed is important), or Robotic Vision/Object Scanning.

Direct Air Capture: A technology-based approach to reversing climate change

Carlos Jimenez Haertel - CTO @Climeworks

Cutting carbon emissions no longer suffices to keep global warming well below the 2° limit agreed in the Paris Climate Accords. Rather, it’s necessary to actively remove CO2 from the atmosphere, up to at least several billion tons annually by 2050. Over the past decade, so-called Direct Air Capture (DAC) methods have come to maturity, which can separate the highly-diluted CO2 from ambient air and store it safely and permanently underground. So far, the development of these technologies has been supported by voluntary carbon removal markets; however, growing DAC to a relevant scale will require putting in place bold climate policies globally. Will we get there in time?

MoBierIO

Stuart Heap - Fullstack Developer @Motius

We build an app outlining available beers from our Motius Bar. The app will allow remote access to the bar and the rating of our own Motius beers (MoBier). In doing so, we will discover the most exciting backend technologies promising to replace Python (i.e fastAPI, rocket, diesel).

Deep-dive in Self-supervised Learning

Daniel Wessel - AI Cluster Lead @Motius

Self-supervised learning is a means for training computers to do tasks without humans providing labeled data. We will discover the latest advancements in this field and develop a blueprint for how to apply this technology in projects related to Computer Vision or NLP.

3D & AR at the Motius Office

Sergej Stankeev - UX & UI Designer @Motius

VR & AR becomes more popular and creates endless opportunities for designers and developers. During Discovery we will deepen our experience with the latest 3D advancements in this area and run several tests by creating an AR Tour through our Motius Office.

Deep-dive in Transformer Architecture

Robin Dehde - Tech Specialist @Motius

Transformers are model architectures that appeared in 2017 and have gained attention with publications by OpenAI, DeepMind, and Google research. With Transformers it is for example possible to let an Artificial intelligence write code or even text in a natural-sounding way. Originally this came at a cost as Transformer architectures used to be very extensive and not very applicable in most cases. With pre-trained architectures, there is an accessible solution for this problem. During Discovery we want to test possible applications for our projects.

Motius Radar

Frederik Zöls - Tech Specialist @Motius

Radar Technology offers a lot of advantages compared to other technologies in privacy-critical applications. We will build a state-of-the-art Radar processing pipeline and eventually provide data to our own Motius IoT System (Motius.io). This would enable a variety of relevant use cases in i.e people counting, crowd control, space observation, or capacity management.

Next Level Reinforcement Learning

Jan Wasgint - Software Engineer @Motius

Reinforcement learning is a subfield of machine learning concerned with how an intelligent agent perceives and interprets his environment, takes actions, and learns through trial and error. We want to explore how multiple agents could learn to work together. Possible applications could be in Robotics, Autonomous driving, Digital Marketing (i.e content recommendation, personalized Ads), or Smart City (i.e intelligent traffic light systems)

Service Mesh With Envoy

Philipp Diller - Software Engineer @Motius

One big trend we see is decentralized Tech. And the majority of operational problems that arise when moving to a distributed architecture are ultimately grounded in two areas: 1. Networking and 2. Observability. Envoy is essentially a modern version of a proxy that can be configured through APIs, based on which many different usage scenarios are derived — such as API Gateway, sidecar proxy in service mesh, and edge proxy. This will allow us to scale and monitor large scale distributed systems.

The Gym for Artificial Training

Justin Iszatt - Tech Specialist @Motius

Training environments offer a great and cost-efficient way for prototyping in many different domains. With this project, we aim to create and use such a setup for Machine Learning.
This will help us train our own Reinforcement Learning Agents,  create artificial training data for ML applications, and might also be used to plan and simulate robotic applications.