Hasil untuk "Fossil man. Human paleontology"

Valentin Nidergas, Jiří Hadrava, Pauline Coster et al.

Here, we describe Intricodon cryptodaemoniacus, a new fossil genus and species of hoverfly (Diptera: Syrphidae), representing the second described hoverfly from the middle Oligocene of Céreste (Luberon, Southern France). It has unique, highly specialized characters like hind legs strongly thickened with processes on femur and tibia and the shape of the wing veins R4+5 and M1. Based on the visible characters on the well preserved compression fossil specimen, the new genus resembles some taxa currently attributed to the Merodontini in the wing venation and/or leg morphology. The systematic affinities and potential palaeoecology of the new genus are discussed through the comparison to the more similar Recent genera.

Robin Fischer, Anton Pichler

Mobilising private capital is a critical bottleneck of the energy transition, yet recent crisis-driven windfall profits for fossil power firms suggest that market signals may still favour carbon-intensive assets. Here we analyse a panel of 900 European power firms (2001-2023) to resolve whether these profits reflect a durable profitability advantage or a crisis-driven anomaly. Using machine-learning clustering and Bayesian model averaging, we identify a structural divergence: wind and solar portfolios exhibit rising profitability, with return on assets among wind-dominated firms increasing by over 6% between 2014 and 2023. Conversely, higher fossil portfolio shares are increasingly associated with lower profitability, with marginal effects reaching -4% by 2023, while renewable-dominated firms match or outperform their fossil-heavy counterparts across most European regions. These findings suggest that the record profits of fossil incumbents were distinct outliers, masking an ongoing decline in the profitability of carbon-intensive business models.

Azucena Solis-Añorve, Mónica R. Buono

The marine outcrops of the Patagonian Miocene (Argentina) encompass one of the most important assemblages of fossil mysticetes recorded worldwide, including key records of extant lineages. The Patagonian Late Miocene records comprise balaenids (right whales) or cetotheriids neobalaenines (pygmy right whales). In the last years, the increase of fieldwork efforts in Miocene outcrops of Patagonia has led to the discovery of new specimens, thus expanding our knowledge of mysticetes diversity from regional and global perspectives. In this work, we describe isolated ear bones collected from the Late Miocene Puerto Madryn Formation 12–2.7 Ma (Serravalian to Piacenzian), Península Valdés (Chubut Province), preliminarily attributed to Cetotheriidae. Although the lack of diagnostic elements precludes a more precise identification (at genus or specific level) within this family, our studies reveal a previously unknown diversity of morphotypes, indicating significant taxonomic diversity among these Patagonian mysticetes. In addition, the South Atlantic cetotheriid assemblage appears to be represented exclusively by large taxa, probably of pelagic habits, which may have facilitated their dispersal into the southern basins. Finally, these records suggest a rapid dispersion of cetotheriids from the Paratethys into the Southern Hemisphere during the Late Miocene, with the Southwest Atlantic Ocean representing their southernmost limit of distribution.

Vaclav Knapp, Matyas Bohacek

Recent pose-transfer methods aim to generate temporally consistent and fully controllable videos of human action where the motion from a reference video is reenacted by a new identity. We evaluate three state-of-the-art pose-transfer methods -- AnimateAnyone, MagicAnimate, and ExAvatar -- by generating videos with actions and identities outside the training distribution and conducting a participant study about the quality of these videos. In a controlled environment of 20 distinct human actions, we find that participants, presented with the pose-transferred videos, correctly identify the desired action only 42.92% of the time. Moreover, the participants find the actions in the generated videos consistent with the reference (source) videos only 36.46% of the time. These results vary by method: participants find the splatting-based ExAvatar more consistent and photorealistic than the diffusion-based AnimateAnyone and MagicAnimate.

Maxence Grand, Damien Pellier, Francis Jambon

The primary objective of the dataset is to provide a better understanding of the coupling between human actions and gaze in a shared working environment with a cobot, with the aim of signifcantly enhancing the effciency and safety of humancobot interactions. More broadly, by linking gaze patterns with physical actions, the dataset offers valuable insights into cognitive processes and attention dynamics in the context of assembly tasks. The proposed dataset contains gaze and action data from approximately 80 participants, recorded during simulated industrial assembly tasks. The tasks were simulated using controlled scenarios in which participants manipulated educational building blocks. Gaze data was collected using two different eye-tracking setups -head-mounted and remote-while participants worked in two positions: sitting and standing.

Xiangrong, Zhu, Yuan Xu et al.

Human cognition is constrained by processing limitations, leading to cognitive overload and inefficiencies in knowledge synthesis and decision-making. Large Language Models (LLMs) present an opportunity for cognitive augmentation, but their current reactive nature limits their real-world applicability. This position paper explores the potential of context-aware cognitive augmentation, where LLMs dynamically adapt to users' cognitive states and task environments to provide appropriate support. Through a think-aloud study in an exhibition setting, we examine how individuals interact with multi-modal information and identify key cognitive challenges in structuring, retrieving, and applying knowledge. Our findings highlight the need for AI-driven cognitive support systems that integrate real-time contextual awareness, personalized reasoning assistance, and socially adaptive interactions. We propose a framework for AI augmentation that seamlessly transitions between real-time cognitive support and post-experience knowledge organization, contributing to the design of more effective human-centered AI systems.

Agnia Sergeyuk, Ilya Zakharov, Ekaterina Koshchenko et al.

The integration of Artificial Intelligence (AI) into Integrated Development Environments (IDEs) is reshaping software development, fundamentally altering how developers interact with their tools. This shift marks the emergence of Human-AI Experience in Integrated Development Environment (in-IDE HAX), a field that explores the evolving dynamics of Human-Computer Interaction in AI-assisted coding environments. Despite rapid adoption, research on in-IDE HAX remains fragmented, which highlights the need for a unified overview of current practices, challenges, and opportunities. To provide a structured overview of existing research, we conduct a systematic literature review of 90 studies, summarizing current findings and outlining areas for further investigation. We organize key insights from reviewed studies into three aspects: Impact, Design, and Quality of AI-based systems inside IDEs. Impact findings show that AI-assisted coding enhances developer productivity but also introduces challenges, such as verification overhead and over-reliance. Design studies show that effective interfaces surface context, provide explanations and transparency of suggestion, and support user control. Quality studies document risks in correctness, maintainability, and security. For future research, priorities include productivity studies, design of assistance, and audit of AI-generated code. The agenda calls for larger and longer evaluations, stronger audit and verification assets, broader coverage across the software life cycle, and adaptive assistance under user control.

Lene L. Delsett, Adam S. Smith, Stephen Ingrams et al.

A plesiosaur specimen collected from Ellesmere Island (Nunavut, Arctic Canada) by Danish geologist Johannes Troelsen in 1952 is described for the first time. The plesiosaur is late Berriasian to early Valanginian in age based on palynostratigraphy. The specimen is the only plesiosaur known from the Lower Cretaceous of the Sverdrup Basin in the Canadian Arctic, and is assigned to the cryptoclidid genus Colymbosaurus. From a taxonomic point of view, the presence of vertebrae from several regions and four propodials improve our understanding of the morphology of the genus. Furthermore, Colymbosaurus is shown to have survived through the Jurassic–Cretaceous transition. Its presence in the Sverdrup Basin is additional evidence for the connectivity of Arctic Canada and the Svalbard region during the Jurassic–Cretaceous transition, at a time when sea levels were low and microplankton, like dinoflagellates, experienced enhanced provincialism. Last but not least, the new plesiosaur adds to our knowledge of the palaeoenvironment of the Sverdrup Basin, ranking at the top of a food chain that is largely unrecorded from the area, due to adverse taphonomy and diagenesis.

Francisco Tello, Luis Libido, Victor Moctezuma

The extinct dung beetle Onthophagus pilauco Tello, Verdú, Rossini & Zunino represents the southernmost paleorecord of the onthophagine dung beetle fauna in the New World. Here, we report new fore-, mid-, and posterior leg fossil remains of this extinct beetle and present a morphological comparison of the leg phenotypes observed in the fossil record. Moreover, we illustrate and describe the fossil legs and provide paleoecological inferences based on morphological analysis. We hypothesized that the differences in size and morphology of the remains are related to sexual dimorphism. To test this hypothesis, we applied parametric and non-parametric statistical approaches to the measurement of leg traits. Based on the leg traits identified in O. pilauco, we also discuss the phylogenetic affinities with modern species. Significant differences in foretibial morphology (i.e., length, width, tooth shape, curvedness) support the sexual dimorphism hypothesis. We found that O. pilauco’s forelegs shared a lack of secondary denticles with the O. hippopotamus species complex, although our diagnostic suggests that this trait evolved independently in these species groups. Finally, we discuss the paleoecology of the species based on the morphology of males and females and propose that this extinct species had paracoprid and hypophagic nesting behavior.

Mariel Ferrari

Systematic knowledge of Early and Middle Jurassic marine gastropods from Argentina has been supplied during the last decade through several contributions. The present paper describes a new marine gastropod fauna for the Jurassic of the Neuquén Basin, represented by four major taxa such as Patellogastropoda, Vetigastropoda, Caenogastropoda and Heterobranchia, including 14 new species of the genera Scurriopsis (Hennocquia), Eucyclus, Pleurotomaria, Laevitomaria, Obornella, Crossostoma, Palaeorissoina, Euconactaeon, Striactaeonina and Sulcoactaeon; members of Ambercyclus and Rhabdocolpus are also described from the studied area. Most of these taxa are first reported in the Jurassic of Argentina, extending their palaeobiogeographical distribution in South America at that time. Particularly, the genera Obornella, Palaeorissoina, and Euconactaeon also extend their chronostratigraphic distribution in the Jurassic marine deposits of the Neuquén Basin, from the late Hettangian (Palaeorissoina) and late Pliensbachian (Obornella) to the Aalenian–Bajocian (Euconactaeon). The new fossils described here expand the known diversity of the marine gastropod faunas in the Jurassic of Argentina and contribute to the paleontological knowledge of the group in the Mesozoic of the Southern Hemisphere.

Faustino Prieto, Catalina B. García-García, Román Salmerón Gómez

Carbon dioxide (CO2) emissions have emerged as a critical issue with profound impacts on the environment, human health, and the global economy. The steady increase in atmospheric CO2 levels, largely due to human activities such as burning fossil fuels and deforestation, has become a major contributor to climate change and its associated catastrophic effects. To tackle this pressing challenge, a coordinated global effort is needed, which necessitates a deep understanding of emissions patterns and trends. In this paper, we explore the use of statistical modelling, specifically the lognormal distribution, as a framework for comprehending and predicting CO2 emissions. We build on prior research that suggests a complex distribution of emissions and seek to test the hypothesis that a simpler distribution can still offer meaningful insights for policy-makers. We utilize data from three comprehensive databases and analyse six candidate distributions (exponential, Fisk, gamma, lognormal, Lomax, Weibull) to identify a suitable model for global fossil CO2 emissions. Our findings highlight the adequacy of the lognormal distribution in characterizing emissions across all countries and years studied. Furthermore, to provide additional support for this distribution, we provide statistical evidence supporting the applicability of Gibrat's law to those CO2 emissions. Finally, we employ the lognormal model to predict emission parameters for the coming years and propose two policies for reducing total fossil CO2 emissions. Our research aims to provide policy-makers with accurate and detailed information to support effective climate change mitigation strategies.

Luca Castri, Gloria Beraldo, Sariah Mghames et al.

Deploying robots in human-shared spaces requires understanding interactions among nearby agents and objects. Modelling cause-and-effect relations through causal inference aids in predicting human behaviours and anticipating robot interventions. However, a critical challenge arises as existing causal discovery methods currently lack an implementation inside the ROS ecosystem, the standard de facto in robotics, hindering effective utilisation in robotics. To address this gap, this paper introduces ROS-Causal, a ROS-based framework for onboard data collection and causal discovery in human-robot spatial interactions. An ad-hoc simulator, integrated with ROS, illustrates the approach's effectiveness, showcasing the robot onboard generation of causal models during data collection. ROS-Causal is available on GitHub: https://github.com/lcastri/roscausal.git.

Charles Ver Straeten

The Devonian strata in New York State were the standard section for North America for over 100 years, and remain a significant reference for regional to global correlation and research. Since publication of L. V. Rickard’s (1975) New York Devonian correlation chart, various higher-resolution stratigraphic analyses have been employed, sometimes at bed-by-bed scale. These include sequence-, bio-, event-, chemo-, and other -stratigraphic approaches, along with increasingly finer-resolution geochronologic dating of airfall volcanic tephras. Results have led to many new interpretations and insights of the succession. The purpose of this three-volume work is to produce a new Devonian stratigraphic synthesis for New York State, and to record, often in detail, current knowledge of the succession, and various other geologic and paleontologic aspects of it for current and future research and discussion. The purpose of this chapter is to provide overviews of the Devonian Period, the Devonian of North America (“Laurentia”), the Devonian of eastern Laurentia, and the Devonian of New York State. Furthermore, this review extends beyond the sedimentary rock and paleobiological record, and beyond the United States, Canada, and northern Mexico, to also summarize aspects of Devonian orogenesis, metasedimentary foreland basin fill, silicic igneous activity, complexities of terranes of Mexico and Central America, and Appalachian faunas that extended into South America. The Devonian Period as a whole encompasses 60 million years of time, approximately 419 to 359 million years ago. During that time, shallow seas covered large continental areas; climate was warmer globally than our current climate, during the late stage of a global greenhouse climate. By the end of the Devonian, that warm climate was descending into a time of global icehouse conditions, with widespread glaciation. The positions of modern continental masses were much different. During the Devonian Period, Life first fully colonized the land, led by primitive spore-bearing plants, small arthropods, and apparently by the Middle Devonian, the first tetrapod (“four-legged”) animals, which evolved from bony fishes. Decimeter-tall plants at the beginning of the period had evolved to tree-size forms by the Middle Devonian, approximately 30 million years later, and Earth’s first forest ecosystems arose. Devonian strata are widespread around the ancient continent “Laurentia,” which approximately corresponds to modern North America). At that time, Laurentia straddled the equator, with New York State and the Appalachian region somewhat north of 30° south latitude. Shallow epicontinental seas covered large but varying amounts of the continent over the period. Mountain belts formed on the eastern, northern, and western margins of Laurentia, due to plate tectonic collisions with smaller continental masses, exotic terranes, and volcanic island arcs. Through the Early to Middle Devonian, seas in western and eastern Laurentia were separated by a “transcontinental arch,” and generally had distinctly different marine faunas. In the latest Middle Devonian, sea level transgressed over the land barrier of the Laurentian Transcontinental Arch and the Canadian Shield, and those marine faunas mixed, leading to a more global cosmopolitan fauna in the Late Devonian. Anomalously, however, Early and Middle Devonian Laurentian shallow marine faunas are found in Devonian rocks in Central and South America, which were part of the southern Gondwana continent, generally thought to be separated from Laurentia by oceanic water depths at that time. During the Devonian, eastern Laurentia was an active tectonic margin, related to continent-continent collisions with various terranes/smaller continental masses. The Caledonian, Acadian, and Neoacadian orogenies resulted in compressional and some transpressional tectonics, and the uplift of an extensive mountain belt from east Greenland to Alabama and Georgia. Crustal loading of the orogen in eastern Laurentia led to subsidence and formation of a retroarc Acadian-Neoacadian Foreland Basin, which was initially filled with marine waters, followed by gradual overfilling to above sea level by massive volumes of synorogenic sediments from the east. The resulting lands were the site of some of the earliest forests on Earth, preserved at several sites in New York State, and forest ecosystems. Large-scale deformation, seismic activity, and metamorphism in the mountain belt were accompanied by igneous processes, including explosive eruption of felsic volcanic ash and other material, collectively termed “tephra,” also sometimes termed ash or tuff layers, or if diagenetically altered, sometimes termed bentonite, K-bentonite, metabentonite, or tonstein layers. These explosive Devonian eruptions sent volcanic tephra high into the atmosphere, and easterly winds spread airfall volcanic “tephra layers” across the eastern United States. Meanwhile, rock decay in the mountains led to the erosion, transport, and deposition of massive volumes of clays, silt, sand, and gravel into the Acadian-Neoacadian Foreland Basin, and beyond. Devonian rocks in New York are found at or just below the surface across approximately 40% of the state (~50,500 km2/19,500 mi2). The strata are generally undeformed and gently dipping, and while often covered by soil, glacial sediments, and vegetative cover, are relatively widely found in natural and man-made exposures. Three relatively thin intervals of carbonates are accompanied by eastward thickening wedges of synorogenic mudrocks, sandstones, and minor conglomerates. The history of geological and paleontological observation and study in New York began in the late 18th century. The first professional geologists appeared in the early 19th century. Since the advent of the first geological survey of New York State in 1836, the Devonian Period (nearly termed the “Erian Period” for New York’s Devonian-age rocks) has been the focus of a great volume of research which continues today. The Devonian succession in New York includes strata from all seven stages of the period, with erosional gaps of small to major significance. In addition to a range of marine facies, nearly one quarter of the entire area of Devonian bedrock in the state was deposited in terrestrial settings, with massive volumes of siliciclastic sediments shed off of Acadian-Neoacadian highlands to the east, that also feature the fossils of Earth’s oldest known forest ecosystems. The stratigraphic philosophy in New York has long evolved toward a hybrid classification, wherein groups, formations, and bed-level units are largely time-rock/allostratigraphic to occasionally chronostratigraphic, with lithostratigraphy often ascribed to member-level divisions (e.g., Pragian to Givetian strata, middle Lower to upper Middle Devonian). However, in some intervals, such as Frasnian strata (lower Upper Devonian), group-level units are time-rock units, and formation-level units within groups are largely lithostratigraphic. Forty-eight years of research since Rickard’s (1975) New York Devonian correlation chart permits development of a new, more refined chart (forthcoming), and also permits a new synthesis of Devonian rocks and fossils in New York, presented in this work of twelve chapters, with additional digital appendices.

FRANCESCO BATTISTA, ANA MARIA RIBEIRO, FERNANDO ERTHAL et al.

One of the main “databases”, on which palaeontologists carry on their studies, is constituted by palaeontological collections. These collections are the final result of fieldwork and surveys, sampling activities, preparation and curatorial processes, and analyses. However, the content of a palaeontological collection can also be strongly biased, leading researchers to post-collection skewed results. Post-collection biases (e.g., breakage, loss of fragments, etc.) are directly linked to human activities, occurring during excavation, transport, preparation, and storage. Here, we present the case of the vertebrate remains from the Brazilian lower Carnian Santacruzodon Assemblage Zone (Santa Cruz Sequence, Santa Maria Supersequence, Paraná Basin). The studied specimens came from the Schoenstatt Sanctuary fossil site, a key outcrop for both the sequence and Santacruzodon AZ. We evaluated vertebrate remains from three Brazilian scientific collections, compiled through more than 25 years of fieldwork. The specimens housed in the three collections present high degrees of post-collection fragmentation, as well as significant differences in the bone elements present, when comparing cranial vs. post-cranial elements. Moreover, some differences in curatorial attitude have also been noticed, especially in restoration choices, leading to “discrimination” in post-collection fossil quality and highlighting the existence of the “craniocentrism” problem.

Comité Organizador RCAPA 2022

Libro de resúmenes de la Reunión de Comunicaciones de la Asociación Paleontológica Argentina, que tuvo lugar en la Universidad Nacional de Salta (Salta, Argentina) del 23 al 25 de noviembre del 2022.

Narla S. Stutz, Patricia Hadler, Francisco R. Negri et al.

The Amazonian region covers a significant part of the South American continent and harbors outstanding biodiversity. However, much of its history is still unknown. This situation has begun to change with paleontological field efforts over the last decades, which have been proving that fossils can be common in this region. Despite their great current species richness and abundance in the area today, marsupials have a sparse fossil record, restricted to a few specimens from handful Cenozoic Amazonian localities. Here we present new records of fossil marsupial teeth from the Solimões Formation (lower Eocene–Pliocene), on the Juruá and Envira riverbanks (Acre, Northwestern Brazil). The localities investigated yield at least four distinct didelphid didelphimorphians at PRE 06 (Ponto Rio Envira: Marmosini ?Marmosa sp., Didelphis cf. D. solimoensis, Thylamys? colombianus, plus unidentified didelphids), and two paucituberculatans from the Juruá River localities (Ponto Rio Juruá: the palaeothentid Palaeothentinae indet. at PRJ 25 and PRJ 33’, and Abderitidae indet. from PRJ 33). In agreement with the associated mammalian faunas, most of the didelphids, except for Thylamys? colombianus from PRE 06, indicate a (?early) Late Miocene age for this locality. Conversely, the abderitid specimens found in situ at PRJ 33 would match a Middle Miocene age. The palaeothentids found at PRJ 25 and PRJ 33’ localities cannot be considered for biostratigraphic inferences, since they were found outside a stratigraphic context. Nevertheless, these paucituberculatans considerably add to our knowledge, as they are the first ever recorded in Brazilian Amazonia.

Jinzhou Lin, Han Gao, Xuxiang Feng et al.

In recent years, the rapid advancement of Large Language Models (LLMs) such as the Generative Pre-trained Transformer (GPT) has attracted increasing attention due to their potential in a variety of practical applications. The application of LLMs with Embodied Intelligence has emerged as a significant area of focus. Among the myriad applications of LLMs, navigation tasks are particularly noteworthy because they demand a deep understanding of the environment and quick, accurate decision-making. LLMs can augment embodied intelligence systems with sophisticated environmental perception and decision-making support, leveraging their robust language and image-processing capabilities. This article offers an exhaustive summary of the symbiosis between LLMs and embodied intelligence with a focus on navigation. It reviews state-of-the-art models, research methodologies, and assesses the advantages and disadvantages of existing embodied navigation models and datasets. Finally, the article elucidates the role of LLMs in embodied intelligence, based on current research, and forecasts future directions in the field. A comprehensive list of studies in this survey is available at https://github.com/Rongtao-Xu/Awesome-LLM-EN.

Diogo Luvizon, Marc Habermann, Vladislav Golyanik et al.

In this work, we consider the problem of estimating the 3D position of multiple humans in a scene as well as their body shape and articulation from a single RGB video recorded with a static camera. In contrast to expensive marker-based or multi-view systems, our lightweight setup is ideal for private users as it enables an affordable 3D motion capture that is easy to install and does not require expert knowledge. To deal with this challenging setting, we leverage recent advances in computer vision using large-scale pre-trained models for a variety of modalities, including 2D body joints, joint angles, normalized disparity maps, and human segmentation masks. Thus, we introduce the first non-linear optimization-based approach that jointly solves for the absolute 3D position of each human, their articulated pose, their individual shapes as well as the scale of the scene. In particular, we estimate the scene depth and person unique scale from normalized disparity predictions using the 2D body joints and joint angles. Given the per-frame scene depth, we reconstruct a point-cloud of the static scene in 3D space. Finally, given the per-frame 3D estimates of the humans and scene point-cloud, we perform a space-time coherent optimization over the video to ensure temporal, spatial and physical plausibility. We evaluate our method on established multi-person 3D human pose benchmarks where we consistently outperform previous methods and we qualitatively demonstrate that our method is robust to in-the-wild conditions including challenging scenes with people of different sizes.

Amir Yazdani, Roya Sabbagh Novin, Andrew Merryweather et al.

3D posture estimation is important in analyzing and improving ergonomics in physical human-robot interaction and reducing the risk of musculoskeletal disorders. Vision-based posture estimation approaches are prone to sensor and model errors, as well as occlusion, while posture estimation solely from the interacting robot's trajectory suffers from ambiguous solutions. To benefit from the advantages of both approaches and improve upon their drawbacks, we introduce a low-cost, non-intrusive, and occlusion-robust multi-sensory 3D postural estimation algorithm in physical human-robot interaction. We use 2D postures from OpenPose over a single camera, and the trajectory of the interacting robot while the human performs a task. We model the problem as a partially-observable dynamical system and we infer the 3D posture via a particle filter. We present our work in teleoperation, but it can be generalized to other applications of physical human-robot interaction. We show that our multi-sensory system resolves human kinematic redundancy better than posture estimation solely using OpenPose or posture estimation solely using the robot's trajectory. This will increase the accuracy of estimated postures compared to the gold-standard motion capture postures. Moreover, our approach also performs better than other single sensory methods when postural assessment using RULA assessment tool.

Christian Diller, Thomas Funkhouser, Angela Dai

We present a generative approach to forecast long-term future human behavior in 3D, requiring only weak supervision from readily available 2D human action data. This is a fundamental task enabling many downstream applications. The required ground-truth data is hard to capture in 3D (mocap suits, expensive setups) but easy to acquire in 2D (simple RGB cameras). Thus, we design our method to only require 2D RGB data at inference time while being able to generate 3D human motion sequences. We use a differentiable 2D projection scheme in an autoregressive manner for weak supervision, and an adversarial loss for 3D regularization. Our method predicts long and complex human behavior sequences (e.g., cooking, assembly) consisting of multiple sub-actions. We tackle this in a semantically hierarchical manner, jointly predicting high-level coarse action labels together with their low-level fine-grained realizations as characteristic 3D human poses. We observe that these two action representations are coupled in nature, and joint prediction benefits both action and pose forecasting. Our experiments demonstrate the complementary nature of joint action and 3D pose prediction: our joint approach outperforms each task treated individually, enables robust longer-term sequence prediction, and improves over alternative approaches to forecast actions and characteristic 3D poses.