Daniel Lenton

I'm a PhD student in The Dyson Robotics Lab at Imperial College London. In my research, I'm exploring the intersection between learning-based geometric reresentations, ego-centric perception, spatial memory, and visuomotor control for robotics.

I am also:

Creator of Ivy, where we're on a mission to unify all Machine Learning (ML) frameworks, join us on our mission and lets-unify.ai!

Co-organizer of The RoundTable Chatroom, where we host events at top AI conferences every year.

Teaching assistant at Imperial College London, having taught courses on robot navigation, algorithms, intro to AI, deep learning and software engineering design.

Ivy  /  Medium  /  LinkedIn  /  Twitter  /  Google Scholar  /  Email

Research and Publications
End-to-End Egospheric Spatial Memory
Daniel lenton, Stephen James, Ronald Clark, Andrew Davison
International Conference on Learning Representations (ICLR), 2021
paper / video / code / project page

ESM encodes the memory in an ego-sphere around the agent, enabling expressive 3D representations. ESM can be trained end-to-end via either imitation or reinforcement learning, and improves both training efficiency and final performance against other memory baselines on visuomotor control tasks.

Ivy: Templated Deep Learning for Inter-Framework Portability
Daniel lenton, Fabio Pardo, Fabian Falck, Stephen James, Ronald Clark
arXiv, 2021
paper / video / code / project page

Ivy is a templated Machine Learning (ML) framework which abstracts existing ML frameworks such that their core functions all exhibit consistent call signatures and syntax. Ivy allows high-level framework-agnostic functions, layers and libraries to be implemented on top.

Unsupervised Path Regression Networks
Michal Pandy, Daniel lenton, Ronald Clark
International Conference on Intelligent Robots and Systems (IROS), 2021
paper / project page

We demonstrate that challenging shortest path problems can be solved via direct spline regression from a neural network, trained in an unsupervised manner without requiring ground truth optimal paths for training. To achieve this, we derive a geometry-dependent optimal cost function whose minima guarantees collision-free solutions.

Waypoint Planning Networks
Alexandru-Iosif Toma, Hussein Ali Jaafar, Hao-Ya Hsueh, Stephen James, Daniel lenton, Ronald Clark, Sajad Saeedi,
International Conference on Computer Vision and Robotics (CVR), 2021
paper / video / code / project page /

Waypoint Planning Networks, or WPN, is a hybrid motion planning algorithm based on LSTMs with a local kernel, a classic algorithm such as A*, and a global kernel using a learned algorithm. WPN produces a more computationally efficient and robust solution than other learned approaches.

MoreFusion: Multi-object Reasoning for 6D Pose Estimation from Volumetric Fusion
Kentaro Wada, Edgar Sucar, Stephen James, Daniel lenton, Andrew Davison
Conference on Computer Vision and Pattern Recognition (CVPR), 2020
paper / video / code / project page

MoreFusion makes 3D object pose proposals from single RGB-D views, accumulates pose estimates and non-parametric occupancy information from multiple views as the camera moves, and performs joint optimization to estimate consistent, non-intersecting poses for multiple objects in contact.